v0.10.0

Output .json file containing information about the program

.github/FUNDING.yml

@@ -0,0 +1,4 @@
+# These are supported funding model platforms
+
+github: [gingerBill]
+patreon: gingerbill

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,39 @@
+---
+name: Bug report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+## Context
+
+Please provide any relevant information about your setup. This is important in case the issue is not reproducible except for under certain conditions.
+
+* Operating System:
+* Odin version/Commit:
+
+## Expected Behavior
+
+Please describe the behavior you are expecting
+
+## Current Behavior
+
+What is the current behavior?
+
+## Failure Information (for bugs)
+
+Please help provide information about the failure if this is a bug. If it is not a bug, please remove the rest of this template.
+
+### Steps to Reproduce
+
+Please provide detailed steps for reproducing the issue.
+
+1. step 1
+2. step 2
+3. you get it...
+
+### Failure Logs
+
+Please include any relevant log snippets or files here.

.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,17 @@
+---
+name: Feature request
+about: Suggest an idea for this project
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Is your feature request related to a problem? Please describe.**
+A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
+
+**Describe the solution you'd like**
+A clear and concise description of what you want to happen.
+
+**Additional context**
+Add any other context or screenshots about the feature request here.

.gitignore

@@ -18,7 +18,7 @@ bld/
 [Bb]in/
 [Oo]bj/
 [Ll]og/
-
+![Cc]ore/[Ll]og/
 # Visual Studio 2015 cache/options directory
 .vs/
 # Uncomment if you have tasks that create the project's static files in wwwroot
@@ -251,7 +251,23 @@ paket-files/
 
 
 # Project Specific
+
+# - Windows
 *.sln
-!misc/llvm-bim/lli.exe
-!misc/llvm-bim/opt.exe
-builds
+builds/
+bin/
+*.exe
+*.obj
+*.pdb
+
+# - Linux/MacOS
+odin
+odin.dSYM
+
+# shared collection
+shared/
+
+# temp files
+* .ll
+*.bc
+*.ll

.travis.yml

@@ -0,0 +1,24 @@
+language: cpp
+git:
+  depth: false
+
+os:
+  - linux
+  - osx
+
+compiler:
+  - clang
+ 
+addons:
+  homebrew:
+    packages:
+    - llvm
+
+script: 
+  - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then export PATH="/usr/local/opt/llvm/bin:$PATH" ; fi
+  - make release
+  - ./odin run examples/demo/demo.odin
+  - ./odin check examples/demo/demo.odin -vet
+
+notifications:
+  email: false

LICENSE

@@ -1,4 +1,4 @@
-Copyright (c) 2016 Ginger Bill. All rights reserved.
+Copyright (c) 2016-2017 Ginger Bill. All rights reserved.
 
 Redistribution and use in source and binary forms, with or without
 modification, are permitted provided that the following conditions are met:

Makefile

@@ -0,0 +1,24 @@
+DISABLED_WARNINGS=-Wno-switch -Wno-pointer-sign -Wno-tautological-constant-out-of-range-compare -Wno-tautological-compare -Wno-macro-redefined -Wno-writable-strings
+LDFLAGS=-pthread -ldl -lm -lstdc++
+CFLAGS=-std=c++11
+CC=clang
+
+OS=$(shell uname)
+
+ifeq ($(OS), Darwin)
+	LDFLAGS:=$(LDFLAGS) -liconv
+endif
+
+all: debug demo
+
+demo:
+	./odin run examples/demo/demo.odin
+
+debug:
+	$(CC) src/main.cpp $(DISABLED_WARNINGS) $(CFLAGS) -g $(LDFLAGS) -o odin
+
+release:
+	$(CC) src/main.cpp $(DISABLED_WARNINGS) $(CFLAGS) -O3 -march=native $(LDFLAGS) -o odin
+
+
+

PROPOSAL-PROCESS.md

@@ -0,0 +1,64 @@
+# The Proposal Process
+
+## Introduction
+
+The Odin project's development process is driven by design and pragmatism. Significant changes to the language, libraries, or tools _must_ be first discussed, and maybe formally documented, before they can be implemented.
+
+This document describes the process for proposing, documenting, and implementing changes to the Odin project.
+
+## The Proposal Process
+
+The proposal process is the process for reviewing a proposal and reaching a decision about whether to accept or decline the proposal.
+
+1. [Ginger Bill](https://github.com/gingerBill) is [BDFL](https://wikipedia.org/wiki/Benevolent_dictator_for_life) and significant changes _must_ be passed by him.
+
+2. The proposal author creates a brief issue describing the proposal.
+
+   Note: There is no need for a design document at this point.<br>
+   Note: A non-proposal issue can be turned into a proposal by simply adding the _proposal_ label.
+
+3. A discussion on the issue tracker will classify the proposal into one of three outcomes:
+	* Accept proposal
+	* Decline proposal
+	* Ask for a design document.
+
+	If the proposal is accepted or declined, the process is done. Otherwise the discussion around the process is expected to identify issues that ought to be addressed in a more detailed design.
+
+4. The proposal author writes a design document to work out details of the proposed design and address the concerns raised in the initial discussion.
+
+5. Once comments and revisions on the design document calm, there is a final discussion on the issue, to reach one of two outcomes:
+	* Accept proposal
+	* Decline proposal
+
+After the proposal is accepted or declined, implementation of the proprosal proceeds in the same way as any other contribution to the project.
+
+## Design Documents
+
+The design document should follow this template:
+
+
+```
+# Proposal: [Title]
+
+Author(s): [Author Name, Co-Author Name]
+Last updated: [Date ISO-8601]
+Discussion at https://github.com/odin-lang/Odin/issues/######
+
+## Abstract
+
+## Background
+
+## Proposal
+
+## Rationale
+
+## Compatibility
+
+## Implementation
+
+```
+
+
+## Help
+
+If you need help with this process, please contact an Odin contributor by posting an issue to the [issue tracker](https://github.com/odin-lang/Odin/issues).

README.md

@@ -1,4 +1,26 @@
-<img src="logo-slim.png" alt="Odin logo" height="74">
+<p align="center">
+    <img src="misc/logo-slim.png" alt="Odin logo" height="120">
+    <br/>
+    A fast, concise, readable, pragmatic and open sourced programming language.
+    <br/>
+    <br/>
+    <a href="https://github.com/odin-lang/odin/releases/latest">
+        <img src="https://img.shields.io/github/release/odin-lang/odin.svg">
+    </a>
+    <a href="https://github.com/odin-lang/odin/releases/latest">
+        <img src="https://img.shields.io/badge/platforms-Windows%20|%20Linux%20|%20macOS-green.svg">
+    </a>
+    <a href="https://github.com/odin-lang/odin/blob/master/LICENSE">
+        <img src="https://img.shields.io/github/license/odin-lang/odin.svg">
+    </a>
+    <br>
+    <a href="https://ci.appveyor.com/project/ThisDrunkDane/odin-vf0ap">
+        <img src="https://ci.appveyor.com/api/projects/status/qss6l921c0eu85u6/branch/master?svg=true">
+    </a>
+    <a href="https://travis-ci.org/odin-lang/Odin">
+        <img src="https://travis-ci.org/odin-lang/Odin.svg?branch=master">
+    </a>
+</p>
 
 # The Odin Programming Language
 
@@ -7,8 +29,127 @@ The Odin programming language is fast, concise, readable, pragmatic and open sou
 * high performance
 * built for modern systems
 * joy of programming
-* metaprogramming
-* designed for good programmers
+
+Website: [https://odin-lang.org/](https://odin-lang.org/)
+
+```go
+package main
+
+import "core:fmt"
+
+main :: proc() {
+	program := "+ + * 😃 - /";
+	accumulator := 0;
+
+	for token in program {
+		switch token {
+		case '+': accumulator += 1;
+		case '-': accumulator -= 1;
+		case '*': accumulator *= 2;
+		case '/': accumulator /= 2;
+		case '😃': accumulator *= accumulator;
+		case: // Ignore everything else
+		}
+	}
+
+	fmt.printf("The program \"%s\" calculates the value %d\n",
+	           program, accumulator);
+}
+
+```
+
+
+## Documentation
+
+#### [Getting Started](https://odin-lang.org/docs/install)
+
+Instructions for downloading and install the Odin compiler and libraries.
+
+### Learning Odin
+
+#### [Overview of Odin](https://odin-lang.org/docs/overview)
+
+An overview of the Odin programming language.
+
+#### [Frequently Asked Questsions (FAQ)](https://odin-lang.org/docs/faq)
+
+Answers to common questsions about Odin.
+
+#### [The Odin Wiki](https://github.com/odin-lang/Odin/wiki)
+
+A wiki maintained by the Odin community.
+
+#### [Odin Discord](https://discord.gg/sVBPHEv)
+
+Get live support and talk with other odiners on the Odin Discord.
+
+### References
+
+#### [Language Specification](https://odin-lang.org/ref/spec)
+
+The official Odin Language specification.
+
+### Articles
+
+#### [The Odin Blog](https://odin-lang.org/blog)
+
+The official blog of the Odin programming language, featuring announcements, news, and in-depth articles by the Odin team and guests.
+
+## Setup
+
+Odin only supports x86-64 at the moment (64-bit), relies on LLVM for code generation and an external linker.
+
+In addition, the following platform-specific steps are necessary:
+
+- Windows
+    * Have Visual Studio installed (MSVC 2010 or later, for the linker)
+    * Have a copy of `opt.exe` and `llc.exe` in `Odin/bin`. Pre-built Windows binaries can be found [here](https://github.com/odin-lang/Odin/releases/tag/llvm-windows) and *must* be explicitly copied
+    * Open a valid command prompt:
+        * **Basic:** run the `x64 Native Tools Command Prompt for VS2017` shortcut bundled with VS 2017, or
+        * **Advanced:** run `vcvarsall.bat x64` from a blank `cmd` session
+
+- MacOS
+    * Have LLVM explicitly installed (`brew install llvm`)
+    * Have XCode installed (version X.X or later, for linking)
+    * Make sure the LLVM binaries and the linker are added to your `$PATH` environmental variable
+
+- GNU/Linux
+    * Have LLVM installed (opt/llc)
+    * Have Clang installed (version X.X or later, for linking)
+    * Make sure the LLVM binaries and the linker are added to your `$PATH` environmental variable
+
+Then build the compiler by calling `build.bat` (Windows) or `make` (Linux/MacOS). This will automatically run the demo program if successful.
+
+**Notes for Linux:**: The compiler currently relies on the `core` and `shared` library collection being relative to the compiler executable. Installing the compiler in the usual sense (to `/usr/local/bin` or similar) is therefore not as straight forward as you need to make sure the mentioned libraries are available. As a result, it is recommended to simply explicitly invoke the compiler with `/path/to/odin` in your preferred build system, or add `/path/to/odin` to `$PATH`.
+
+Please read the [Getting Started Guide](https://github.com/odin-lang/Odin/wiki#getting-started-with-odin) for more information.
+
+## Requirements to build and run
+
+Please read the [Getting Started Guide](https://github.com/odin-lang/Odin/wiki#getting-started-with-odin).
+
+- Windows
+	* x86-64
+	* MSVC 2010 installed (C++11 support)
+	* [LLVM binaries](https://github.com/odin-lang/Odin/releases/tag/llvm-windows) for `opt.exe`, `llc.exe`, and `lld-link.exe`
+	* Requires MSVC's link.exe as the linker
+		* run `vcvarsall.bat` to setup the path
+
+- MacOS
+	* x86-64
+	* LLVM explicitly installed (`brew install llvm`)
+	* XCode installed (for the linker)
+
+- GNU/Linux
+	* x86-64
+	* Build tools (ld)
+	* LLVM installed
+	* Clang installed (temporary - this is Calling the linker for now)
+
+## Warnings
+
+* This is still highly in development and the language's design is quite volatile.
+* Syntax is not fixed.
 
 ## Demonstrations:
 * First Talk & Demo
@@ -18,38 +159,11 @@ The Odin programming language is fast, concise, readable, pragmatic and open sou
 * [Composition & Refactorability](https://www.youtube.com/watch?v=n1wemZfcbXM)
 * [Introspection, Modules, and Record Layout](https://www.youtube.com/watch?v=UFq8rhWhx4s)
 * [push_allocator & Minimal Dependency Building](https://www.youtube.com/watch?v=f_LGVOAMb78)
-* [when, for, & procedure overloading](https://www.youtube.com/watch?v=OzeOekzyZK8)
+* [when, for & procedure overloading](https://www.youtube.com/watch?v=OzeOekzyZK8)
+* [Context Types, Unexported Entities, Labelled Branches](https://www.youtube.com/watch?v=CkHVwT1Qk-g)
+* [Bit Fields, i128 & u128, Syntax Changes](https://www.youtube.com/watch?v=NlTutcLyF64)
+* [Default and Named Arguments; Explicit Parametric Polymorphism](https://www.youtube.com/watch?v=-XQZE6S6zUU)
+* [Loadsachanges](https://www.youtube.com/watch?v=ar0vFMoMtrI)
+* [Packages, Bit Sets, cstring](https://youtu.be/b8bJbjiXZrQ)
+	- [Q&A](https://youtu.be/5jmxyIfyyTk)
 
-## Requirements to build and run
-
-* Windows
-* x86-64
-* MSVC 2015 installed (C99 support)
-* Requires MSVC's link.exe as the linker
-	- run `vcvarsall.bat` to setup the path
-
-## Warnings
-
-* This is still highly in development and the language's design is quite volatile.
-* Syntax is not fixed.
-
-## Roadmap
-
-Not in any particular order and not be implemented
-
-* Compile Time Execution (CTE)
-	- More metaprogramming madness
-	- Compiler as a library
-	- AST inspection and modification
-* CTE-based build system
-* Replace LLVM backend with my own custom backend
-* Improve SSA design to accommodate for lowering to a "bytecode"
-* SSA optimizations
-* Documentation Generator for "Entities"
-* Multiple Architecture support
-* Debug Information
-	- pdb format too
-* Command Line Tooling
-* Compiler Internals:
-	- Big numbers library
-	- Multithreading for performance increase

appveyor.yml

@@ -0,0 +1,19 @@
+image: 
+  - Visual Studio 2017
+shallow_clone: true
+
+platform: x64
+
+install:
+  - cd bin
+  - appveyor DownloadFile https://github.com/odin-lang/Odin/releases/download/llvm-windows/llvm-binaries.zip
+  - 7z x llvm-binaries.zip > nul
+  - cd ..
+
+build_script: 
+  - call "C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Auxiliary\Build\vcvars64.bat"
+  - ./build_ci.bat
+
+test_script:
+  - odin run examples/demo/demo.odin
+  - odin check examples/demo/demo.odin -vet

bin/CODE_OWNERS.TXT

@@ -1,200 +0,0 @@
-This file is a list of the people responsible for ensuring that patches for a
-particular part of LLVM are reviewed, either by themself or by someone else.
-They are also the gatekeepers for their part of LLVM, with the final word on
-what goes in or not.
-
-The list is sorted by surname and formatted to allow easy grepping and
-beautification by scripts.  The fields are: name (N), email (E), web-address
-(W), PGP key ID and fingerprint (P), description (D), and snail-mail address
-(S). Each entry should contain at least the (N), (E) and (D) fields.
-
-N: Joe Abbey
-E: jabbey@arxan.com
-D: LLVM Bitcode (lib/Bitcode/* include/llvm/Bitcode/*)
-
-N: Owen Anderson
-E: resistor@mac.com
-D: SelectionDAG (lib/CodeGen/SelectionDAG/*)
-
-N: Rafael Avila de Espindola
-E: rafael.espindola@gmail.com
-D: Gold plugin (tools/gold/*)
-
-N: Justin Bogner
-E: mail@justinbogner.com
-D: InstrProfiling and related parts of ProfileData
-
-N: Chandler Carruth
-E: chandlerc@gmail.com
-E: chandlerc@google.com
-D: Config, ADT, Support, inlining & related passes, SROA/mem2reg & related passes, CMake, library layering
-
-N: Evan Cheng
-E: evan.cheng@apple.com
-D: parts of code generator not covered by someone else
-
-N: Eric Christopher
-E: echristo@gmail.com
-D: Debug Information, autotools/configure/make build, inline assembly
-
-N: Greg Clayton
-E: gclayton@apple.com
-D: LLDB
-
-N: Marshall Clow
-E: mclow.lists@gmail.com
-D: libc++
-
-N: Peter Collingbourne
-E: peter@pcc.me.uk
-D: llgo
-
-N: Quentin Colombet
-E: qcolombet@apple.com
-D: Register allocators
-
-N: Duncan P. N. Exon Smith
-E: dexonsmith@apple.com
-D: Branch weights and BlockFrequencyInfo
-
-N: Hal Finkel
-E: hfinkel@anl.gov
-D: BBVectorize, the loop reroller, alias analysis and the PowerPC target
-
-N: Dan Gohman
-E: sunfish@mozilla.com
-D: WebAssembly Backend (lib/Target/WebAssembly/*)
-
-N: Renato Golin
-E: renato.golin@linaro.org
-D: ARM Linux support
-
-N: Venkatraman Govindaraju
-E: venkatra@cs.wisc.edu
-D: Sparc Backend (lib/Target/Sparc/*)
-
-N: Tobias Grosser
-E: tobias@grosser.es
-D: Polly
-
-N: James Grosbach
-E: grosbach@apple.com
-D: MC layer
-
-N: Justin Holewinski
-E: jholewinski@nvidia.com
-D: NVPTX Target (lib/Target/NVPTX/*)
-
-N: Lang Hames
-E: lhames@gmail.com
-D: MCJIT, RuntimeDyld and JIT event listeners
-
-N: Galina Kistanova
-E: gkistanova@gmail.com
-D: LLVM Buildbot
-
-N: Anton Korobeynikov
-E: anton@korobeynikov.info
-D: Exception handling, Windows codegen, ARM EABI
-
-N: Benjamin Kramer
-E: benny.kra@gmail.com
-D: DWARF Parser
-
-N: Sergei Larin
-E: slarin@codeaurora.org
-D: VLIW Instruction Scheduling, Packetization
-
-N: Chris Lattner
-E: sabre@nondot.org
-W: http://nondot.org/~sabre/
-D: Everything not covered by someone else
-
-N: David Majnemer
-E: david.majnemer@gmail.com
-D: IR Constant Folder, InstCombine
-
-N: Dylan McKay
-E: dylanmckay34@gmail.com
-D: AVR Backend
-
-N: Tim Northover
-E: t.p.northover@gmail.com
-D: AArch64 backend, misc ARM backend
-
-N: Diego Novillo
-E: dnovillo@google.com
-D: SampleProfile and related parts of ProfileData
-
-N: Jakob Olesen
-E: stoklund@2pi.dk
-D: TableGen
-
-N: Richard Osborne
-E: richard@xmos.com
-D: XCore Backend
-
-N: Krzysztof Parzyszek
-E: kparzysz@codeaurora.org
-D: Hexagon Backend
-
-N: Paul Robinson
-E: paul_robinson@playstation.sony.com
-D: Sony PlayStation®4 support
-
-N: Chad Rosier
-E: mcrosier@codeaurora.org
-D: Fast-Isel
-
-N: Nadav Rotem
-E: nrotem@apple.com
-D: X86 Backend, Loop Vectorizer
-
-N: Daniel Sanders
-E: daniel.sanders@imgtec.com
-D: MIPS Backend (lib/Target/Mips/*)
-
-N: Duncan Sands
-E: baldrick@free.fr
-D: DragonEgg
-
-N: Kostya Serebryany
-E: kcc@google.com
-D: AddressSanitizer, ThreadSanitizer (LLVM parts)
-
-N: Michael Spencer
-E: bigcheesegs@gmail.com
-D: Windows parts of Support, Object, ar, nm, objdump, ranlib, size
-
-N: Alexei Starovoitov
-E: alexei.starovoitov@gmail.com
-D: BPF backend
-
-N: Tom Stellard
-E: thomas.stellard@amd.com
-E: mesa-dev@lists.freedesktop.org
-D: Release manager for the 3.5 and 3.6 branches, R600 Backend, libclc
-
-N: Evgeniy Stepanov
-E: eugenis@google.com
-D: MemorySanitizer (LLVM part)
-
-N: Andrew Trick
-E: atrick@apple.com
-D: IndVar Simplify, Loop Strength Reduction, Instruction Scheduling
-
-N: Ulrich Weigand
-E: uweigand@de.ibm.com
-D: SystemZ Backend
-
-N: Bill Wendling
-E: isanbard@gmail.com
-D: libLTO, IR Linker
-
-N: Peter Zotov
-E: whitequark@whitequark.org
-D: OCaml bindings
-
-N: Andrey Churbanov
-E: andrey.churbanov@intel.com
-D: OpenMP runtime library

bin/CREDITS.TXT

@@ -1,467 +0,0 @@
-This file is a partial list of people who have contributed to the LLVM
-project.  If you have contributed a patch or made some other contribution to
-LLVM, please submit a patch to this file to add yourself, and it will be
-done!
-
-The list is sorted by surname and formatted to allow easy grepping and
-beautification by scripts.  The fields are: name (N), email (E), web-address
-(W), PGP key ID and fingerprint (P), description (D), snail-mail address
-(S), and (I) IRC handle.
-
-
-N: Vikram Adve
-E: vadve@cs.uiuc.edu
-W: http://www.cs.uiuc.edu/~vadve/
-D: The Sparc64 backend, provider of much wisdom, and motivator for LLVM
-
-N: Owen Anderson
-E: resistor@mac.com
-D: LCSSA pass and related LoopUnswitch work
-D: GVNPRE pass, DataLayout refactoring, random improvements
-
-N: Henrik Bach
-D: MingW Win32 API portability layer
-
-N: Aaron Ballman
-E: aaron@aaronballman.com
-D: __declspec attributes, Windows support, general bug fixing
-
-N: Nate Begeman
-E: natebegeman@mac.com
-D: PowerPC backend developer
-D: Target-independent code generator and analysis improvements
-
-N: Daniel Berlin
-E: dberlin@dberlin.org
-D: ET-Forest implementation.
-D: Sparse bitmap
-
-N: David Blaikie
-E: dblaikie@gmail.com
-D: General bug fixing/fit & finish, mostly in Clang
-
-N: Neil Booth
-E: neil@daikokuya.co.uk
-D: APFloat implementation.
-
-N: Misha Brukman
-E: brukman+llvm@uiuc.edu
-W: http://misha.brukman.net
-D: Portions of X86 and Sparc JIT compilers, PowerPC backend
-D: Incremental bitcode loader
-
-N: Cameron Buschardt
-E: buschard@uiuc.edu
-D: The `mem2reg' pass - promotes values stored in memory to registers
-
-N: Brendon Cahoon
-E: bcahoon@codeaurora.org
-D: Loop unrolling with run-time trip counts.
-
-N: Chandler Carruth
-E: chandlerc@gmail.com
-E: chandlerc@google.com
-D: Hashing algorithms and interfaces
-D: Inline cost analysis
-D: Machine block placement pass
-D: SROA
-
-N: Casey Carter
-E: ccarter@uiuc.edu
-D: Fixes to the Reassociation pass, various improvement patches
-
-N: Evan Cheng
-E: evan.cheng@apple.com
-D: ARM and X86 backends
-D: Instruction scheduler improvements
-D: Register allocator improvements
-D: Loop optimizer improvements
-D: Target-independent code generator improvements
-
-N: Dan Villiom Podlaski Christiansen
-E: danchr@gmail.com
-E: danchr@cs.au.dk
-W: http://villiom.dk
-D: LLVM Makefile improvements
-D: Clang diagnostic & driver tweaks
-S: Aarhus, Denmark
-
-N: Jeff Cohen
-E: jeffc@jolt-lang.org
-W: http://jolt-lang.org
-D: Native Win32 API portability layer
-
-N: John T. Criswell
-E: criswell@uiuc.edu
-D: Original Autoconf support, documentation improvements, bug fixes
-
-N: Anshuman Dasgupta
-E: adasgupt@codeaurora.org
-D: Deterministic finite automaton based infrastructure for VLIW packetization
-
-N: Stefanus Du Toit
-E: stefanus.du.toit@intel.com
-D: Bug fixes and minor improvements
-
-N: Rafael Avila de Espindola
-E: rafael.espindola@gmail.com
-D: The ARM backend
-
-N: Dave Estes
-E: cestes@codeaurora.org
-D: AArch64 machine description for Cortex-A53
-
-N: Alkis Evlogimenos
-E: alkis@evlogimenos.com
-D: Linear scan register allocator, many codegen improvements, Java frontend
-
-N: Hal Finkel
-E: hfinkel@anl.gov
-D: Basic-block autovectorization, PowerPC backend improvements
-
-N: Eric Fiselier
-E: eric@efcs.ca
-D: LIT patches and documentation.
-
-N: Ryan Flynn
-E: pizza@parseerror.com
-D: Miscellaneous bug fixes
-
-N: Brian Gaeke
-E: gaeke@uiuc.edu
-W: http://www.students.uiuc.edu/~gaeke/
-D: Portions of X86 static and JIT compilers; initial SparcV8 backend
-D: Dynamic trace optimizer
-D: FreeBSD/X86 compatibility fixes, the llvm-nm tool
-
-N: Nicolas Geoffray
-E: nicolas.geoffray@lip6.fr
-W: http://www-src.lip6.fr/homepages/Nicolas.Geoffray/
-D: PPC backend fixes for Linux
-
-N: Louis Gerbarg
-E: lgg@apple.com
-D: Portions of the PowerPC backend
-
-N: Saem Ghani
-E: saemghani@gmail.com
-D: Callgraph class cleanups
-
-N: Mikhail Glushenkov
-E: foldr@codedgers.com
-D: Author of llvmc2
-
-N: Dan Gohman
-E: sunfish@mozilla.com
-D: Miscellaneous bug fixes
-D: WebAssembly Backend
-
-N: David Goodwin
-E: david@goodwinz.net
-D: Thumb-2 code generator
-
-N: David Greene
-E: greened@obbligato.org
-D: Miscellaneous bug fixes
-D: Register allocation refactoring
-
-N: Gabor Greif
-E: ggreif@gmail.com
-D: Improvements for space efficiency
-
-N: James Grosbach
-E: grosbach@apple.com
-I: grosbach
-D: SjLj exception handling support
-D: General fixes and improvements for the ARM back-end
-D: MCJIT
-D: ARM integrated assembler and assembly parser
-D: Led effort for the backend formerly known as ARM64
-
-N: Lang Hames
-E: lhames@gmail.com
-D: PBQP-based register allocator
-
-N: Gordon Henriksen
-E: gordonhenriksen@mac.com
-D: Pluggable GC support
-D: C interface
-D: Ocaml bindings
-
-N: Raul Fernandes Herbster
-E: raul@dsc.ufcg.edu.br
-D: JIT support for ARM
-
-N: Paolo Invernizzi
-E: arathorn@fastwebnet.it
-D: Visual C++ compatibility fixes
-
-N: Patrick Jenkins
-E: patjenk@wam.umd.edu
-D: Nightly Tester
-
-N: Dale Johannesen
-E: dalej@apple.com
-D: ARM constant islands improvements
-D: Tail merging improvements
-D: Rewrite X87 back end
-D: Use APFloat for floating point constants widely throughout compiler
-D: Implement X87 long double
-
-N: Brad Jones
-E: kungfoomaster@nondot.org
-D: Support for packed types
-
-N: Rod Kay
-E: rkay@auroraux.org
-D: Author of LLVM Ada bindings
-
-N: Eric Kidd
-W: http://randomhacks.net/
-D: llvm-config script
-
-N: Anton Korobeynikov
-E: asl@math.spbu.ru
-D: Mingw32 fixes, cross-compiling support, stdcall/fastcall calling conv.
-D: x86/linux PIC codegen, aliases, regparm/visibility attributes
-D: Switch lowering refactoring
-
-N: Sumant Kowshik
-E: kowshik@uiuc.edu
-D: Author of the original C backend
-
-N: Benjamin Kramer
-E: benny.kra@gmail.com
-D: Miscellaneous bug fixes
-
-N: Sundeep Kushwaha
-E: sundeepk@codeaurora.org
-D: Implemented DFA-based target independent VLIW packetizer
-
-N: Christopher Lamb
-E: christopher.lamb@gmail.com
-D: aligned load/store support, parts of noalias and restrict support
-D: vreg subreg infrastructure, X86 codegen improvements based on subregs
-D: address spaces
-
-N: Jim Laskey
-E: jlaskey@apple.com
-D: Improvements to the PPC backend, instruction scheduling
-D: Debug and Dwarf implementation
-D: Auto upgrade mangler
-D: llvm-gcc4 svn wrangler
-
-N: Chris Lattner
-E: sabre@nondot.org
-W: http://nondot.org/~sabre/
-D: Primary architect of LLVM
-
-N: Tanya Lattner (Tanya Brethour)
-E: tonic@nondot.org
-W: http://nondot.org/~tonic/
-D: The initial llvm-ar tool, converted regression testsuite to dejagnu
-D: Modulo scheduling in the SparcV9 backend
-D: Release manager (1.7+)
-
-N: Sylvestre Ledru
-E: sylvestre@debian.org
-W: http://sylvestre.ledru.info/
-W: http://llvm.org/apt/
-D: Debian and Ubuntu packaging
-D: Continuous integration with jenkins
-
-N: Andrew Lenharth
-E: alenhar2@cs.uiuc.edu
-W: http://www.lenharth.org/~andrewl/
-D: Alpha backend
-D: Sampling based profiling
-
-N: Nick Lewycky
-E: nicholas@mxc.ca
-D: PredicateSimplifier pass
-
-N: Tony Linthicum, et. al.
-E: tlinth@codeaurora.org
-D: Backend for Qualcomm's Hexagon VLIW processor.
-
-N: Bruno Cardoso Lopes
-E: bruno.cardoso@gmail.com
-I: bruno
-W: http://brunocardoso.cc
-D: Mips backend
-D: Random ARM integrated assembler and assembly parser improvements
-D: General X86 AVX1 support
-
-N: Duraid Madina
-E: duraid@octopus.com.au
-W: http://kinoko.c.u-tokyo.ac.jp/~duraid/
-D: IA64 backend, BigBlock register allocator
-
-N: John McCall
-E: rjmccall@apple.com
-D: Clang semantic analysis and IR generation
-
-N: Michael McCracken
-E: michael.mccracken@gmail.com
-D: Line number support for llvmgcc
-
-N: Vladimir Merzliakov
-E: wanderer@rsu.ru
-D: Test suite fixes for FreeBSD
-
-N: Scott Michel
-E: scottm@aero.org
-D: Added STI Cell SPU backend.
-
-N: Kai Nacke
-E: kai@redstar.de
-D: Support for implicit TLS model used with MS VC runtime
-D: Dumping of Win64 EH structures
-
-N: Takumi Nakamura
-E: geek4civic@gmail.com
-E: chapuni@hf.rim.or.jp
-D: Cygwin and MinGW support.
-D: Win32 tweaks.
-S: Yokohama, Japan
-
-N: Edward O'Callaghan
-E: eocallaghan@auroraux.org
-W: http://www.auroraux.org
-D: Add Clang support with various other improvements to utils/NewNightlyTest.pl
-D: Fix and maintain Solaris & AuroraUX support for llvm, various build warnings
-D: and error clean ups.
-
-N: Morten Ofstad
-E: morten@hue.no
-D: Visual C++ compatibility fixes
-
-N: Jakob Stoklund Olesen
-E: stoklund@2pi.dk
-D: Machine code verifier
-D: Blackfin backend
-D: Fast register allocator
-D: Greedy register allocator
-
-N: Richard Osborne
-E: richard@xmos.com
-D: XCore backend
-
-N: Devang Patel
-E: dpatel@apple.com
-D: LTO tool, PassManager rewrite, Loop Pass Manager, Loop Rotate
-D: GCC PCH Integration (llvm-gcc), llvm-gcc improvements
-D: Optimizer improvements, Loop Index Split
-
-N: Ana Pazos
-E: apazos@codeaurora.org
-D: Fixes and improvements to the AArch64 backend
-
-N: Wesley Peck
-E: peckw@wesleypeck.com
-W: http://wesleypeck.com/
-D: MicroBlaze backend
-
-N: Francois Pichet
-E: pichet2000@gmail.com
-D: MSVC support
-
-N: Vladimir Prus
-W: http://vladimir_prus.blogspot.com
-E: ghost@cs.msu.su
-D: Made inst_iterator behave like a proper iterator, LowerConstantExprs pass
-
-N: Kalle Raiskila
-E: kalle.rasikila@nokia.com
-D: Some bugfixes to CellSPU
-
-N: Xerxes Ranby
-E: xerxes@zafena.se
-D: Cmake dependency chain and various bug fixes
-
-N: Alex Rosenberg
-E: alexr@leftfield.org
-I: arosenberg
-D: ARM calling conventions rewrite, hard float support
-
-N: Chad Rosier
-E: mcrosier@codeaurora.org
-I: mcrosier
-D: AArch64 fast instruction selection pass
-D: Fixes and improvements to the ARM fast-isel pass
-D: Fixes and improvements to the AArch64 backend
-
-N: Nadav Rotem
-E: nrotem@apple.com
-D: X86 code generation improvements, Loop Vectorizer.
-
-N: Roman Samoilov
-E: roman@codedgers.com
-D: MSIL backend
-
-N: Duncan Sands
-E: baldrick@free.fr
-I: baldrick
-D: Ada support in llvm-gcc
-D: Dragonegg plugin
-D: Exception handling improvements
-D: Type legalizer rewrite
-
-N: Ruchira Sasanka
-E: sasanka@uiuc.edu
-D: Graph coloring register allocator for the Sparc64 backend
-
-N: Arnold Schwaighofer
-E: arnold.schwaighofer@gmail.com
-D: Tail call optimization for the x86 backend
-
-N: Shantonu Sen
-E: ssen@apple.com
-D: Miscellaneous bug fixes
-
-N: Anand Shukla
-E: ashukla@cs.uiuc.edu
-D: The `paths' pass
-
-N: Michael J. Spencer
-E: bigcheesegs@gmail.com
-D: Shepherding Windows COFF support into MC.
-D: Lots of Windows stuff.
-
-N: Reid Spencer
-E: rspencer@reidspencer.com
-W: http://reidspencer.com/
-D: Lots of stuff, see: http://wiki.llvm.org/index.php/User:Reid
-
-N: Alp Toker
-E: alp@nuanti.com
-W: http://atoker.com/
-D: C++ frontend next generation standards implementation
-
-N: Craig Topper
-E: craig.topper@gmail.com
-D: X86 codegen and disassembler improvements. AVX2 support.
-
-N: Edwin Torok
-E: edwintorok@gmail.com
-D: Miscellaneous bug fixes
-
-N: Adam Treat
-E: manyoso@yahoo.com
-D: C++ bugs filed, and C++ front-end bug fixes.
-
-N: Lauro Ramos Venancio
-E: lauro.venancio@indt.org.br
-D: ARM backend improvements
-D: Thread Local Storage implementation
-
-N: Bill Wendling
-I: wendling
-E: isanbard@gmail.com
-D: Release manager, IR Linker, LTO
-D: Bunches of stuff
-
-N: Bob Wilson
-E: bob.wilson@acm.org
-D: Advanced SIMD (NEON) support in the ARM backend.
-

bin/LICENSE.TXT

@@ -1,70 +0,0 @@
-==============================================================================
-LLVM Release License
-==============================================================================
-University of Illinois/NCSA
-Open Source License
-
-Copyright (c) 2003-2015 University of Illinois at Urbana-Champaign.
-All rights reserved.
-
-Developed by:
-
-    LLVM Team
-
-    University of Illinois at Urbana-Champaign
-
-    http://llvm.org
-
-Permission is hereby granted, free of charge, to any person obtaining a copy of
-this software and associated documentation files (the "Software"), to deal with
-the Software without restriction, including without limitation the rights to
-use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
-of the Software, and to permit persons to whom the Software is furnished to do
-so, subject to the following conditions:
-
-    * Redistributions of source code must retain the above copyright notice,
-      this list of conditions and the following disclaimers.
-
-    * Redistributions in binary form must reproduce the above copyright notice,
-      this list of conditions and the following disclaimers in the
-      documentation and/or other materials provided with the distribution.
-
-    * Neither the names of the LLVM Team, University of Illinois at
-      Urbana-Champaign, nor the names of its contributors may be used to
-      endorse or promote products derived from this Software without specific
-      prior written permission.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
-FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
-CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS WITH THE
-SOFTWARE.
-
-==============================================================================
-Copyrights and Licenses for Third Party Software Distributed with LLVM:
-==============================================================================
-The LLVM software contains code written by third parties.  Such software will
-have its own individual LICENSE.TXT file in the directory in which it appears.
-This file will describe the copyrights, license, and restrictions which apply
-to that code.
-
-The disclaimer of warranty in the University of Illinois Open Source License
-applies to all code in the LLVM Distribution, and nothing in any of the
-other licenses gives permission to use the names of the LLVM Team or the
-University of Illinois to endorse or promote products derived from this
-Software.
-
-The following pieces of software have additional or alternate copyrights,
-licenses, and/or restrictions:
-
-Program             Directory
--------             ---------
-Autoconf            llvm/autoconf
-                    llvm/projects/ModuleMaker/autoconf
-Google Test         llvm/utils/unittest/googletest
-OpenBSD regex       llvm/lib/Support/{reg*, COPYRIGHT.regex}
-pyyaml tests        llvm/test/YAMLParser/{*.data, LICENSE.TXT}
-ARM contributions   llvm/lib/Target/ARM/LICENSE.TXT
-md5 contributions   llvm/lib/Support/MD5.cpp llvm/include/llvm/Support/MD5.h

bin/README.md

@@ -0,0 +1,30 @@
+# The Odin Programming Language
+
+## Setup
+
+Odin only supports x86-64 at the moment (64-bit), relies on LLVM for code generation and an external linker.
+
+In addition, the following platform-specific steps are necessary:
+
+- Windows
+    * Have Visual Studio installed (MSVC 2010 or later, for the linker)
+    * Have a copy of `opt.exe` and `llc.exe` in `Odin/bin`. Pre-built Windows binaries can be found [here](https://github.com/odin-lang/Odin/releases/tag/llvm-windows) and *must* be explicitly copied
+    * Open a valid command prompt:
+        * **Basic:** run the `x64 Native Tools Command Prompt for VS2017` shortcut bundled with VS 2017, or
+        * **Advanced:** run `vcvarsall.bat x64` from a blank `cmd` session
+
+- MacOS
+    * Have LLVM explicitly installed (`brew install llvm`)
+    * Have XCode installed (version X.X or later, for linking)
+    * Make sure the LLVM binaries and the linker are added to your `$PATH` environmental variable
+
+- GNU/Linux
+    * Have LLVM installed (opt/llc)
+    * Have Clang installed (version X.X or later, for linking)
+    * Make sure the LLVM binaries and the linker are added to your `$PATH` environmental variable
+
+Then build the compiler by calling `build.bat` (Windows) or `make` (Linux/MacOS). This will automatically run the demo program if successful.
+
+**Notes for Linux:**: The compiler currently relies on the `core` and `shared` library collection being relative to the compiler executable. Installing the compiler in the usual sense (to `/usr/local/bin` or similar) is therefore not as straight forward as you need to make sure the mentioned libraries are available. As a result, it is recommended to simply explicitly invoke the compiler with `/path/to/odin` in your preferred build system, or add `/path/to/odin` to `$PATH`.
+
+Please read the [Getting Started Guide](https://github.com/odin-lang/Odin/wiki#getting-started-with-odin) for more information.

build.bat

@@ -4,28 +4,26 @@
 set exe_name=odin.exe
 
 :: Debug = 0, Release = 1
-set release_mode=1
-set compiler_flags= -nologo -Oi -TC -fp:fast -fp:except- -Gm- -MP -FC -GS- -EHsc- -GR-
+set release_mode=0
+set compiler_flags= -nologo -Oi -TP -fp:precise -Gm- -MP -FC -GS- -EHsc- -GR-
 
 if %release_mode% EQU 0 ( rem Debug
 	set compiler_flags=%compiler_flags% -Od -MDd -Z7
 	rem -DDISPLAY_TIMING
 ) else ( rem Release
-	set compiler_flags=%compiler_flags% -O2 -MT -Z7
+	set compiler_flags=%compiler_flags% -O2 -MT -Z7 -DNO_ARRAY_BOUNDS_CHECK
 )
 
 set compiler_warnings= ^
 	-W4 -WX ^
 	-wd4100 -wd4101 -wd4127 -wd4189 ^
-	-wd4201 -wd4204 -wd4244 ^
-	-wd4306 ^
+	-wd4201 -wd4204 ^
 	-wd4456 -wd4457 -wd4480 ^
-	-wd4505 -wd4512 -wd4550
+	-wd4512
 
 set compiler_includes=
 set libs= ^
 	kernel32.lib
-	rem "src\dyncall\lib\*.lib"
 
 set linker_flags= -incremental:no -opt:ref -subsystem:console
 
@@ -38,21 +36,14 @@ if %release_mode% EQU 0 ( rem Debug
 set compiler_settings=%compiler_includes% %compiler_flags% %compiler_warnings%
 set linker_settings=%libs% %linker_flags%
 
-
 del *.pdb > NUL 2> NUL
 del *.ilk > NUL 2> NUL
 
-cl %compiler_settings% "src\main.c" ^
-	/link %linker_settings% -OUT:%exe_name% ^
-	&& odin run code/demo.odin
-	rem && odin build_dll code/example.odin ^
-	rem odin run code/demo.odin
 
-rem pushd src\asm
-rem 	nasm hellope.asm -fwin64 -o hellope.obj ^
-rem 	&& cl /nologo hellope.obj /link kernel32.lib /entry:main  ^
-rem 	&& hellope.exe
-rem popd
+cl %compiler_settings% "src\main.cpp" ^
+	/link %linker_settings% -OUT:%exe_name% ^
+	&& odin run examples/demo/demo.odin
+
+del *.obj > NUL 2> NUL
 
 :end_of_build
-

build.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+
+release_mode=$1
+
+warnings_to_disable="-std=c++11 -Wno-switch -Wno-pointer-sign -Wno-tautological-constant-out-of-range-compare -Wno-tautological-compare -Wno-macro-redefined -Wno-writable-strings"
+libraries="-pthread -ldl -lm -lstdc++"
+other_args=""
+compiler="clang"
+
+if [ -z "$release_mode" ]; then release_mode="0"; fi
+
+if [ "$release_mode" -eq "0" ]; then
+	other_args="${other_args} -g"
+fi
+if [ "$release_mode" -eq "1" ]; then
+	other_args="${other_args} -O3 -march=native"
+fi
+
+if [[ "$(uname)" == "Darwin" ]]; then
+
+	# Set compiler to clang on MacOS
+	# MacOS provides a symlink to clang called gcc, but it's nice to be explicit here.
+	compiler="clang"
+
+	other_args="${other_args} -liconv"
+fi
+
+${compiler} src/main.cpp ${warnings_to_disable} ${libraries} ${other_args} -o odin && ./odin run examples/demo/demo.odin

build_ci.bat

@@ -0,0 +1,24 @@
+@echo off
+
+set exe_name=odin.exe
+
+set compiler_flags= -nologo -Oi -TP -fp:precise -Gm- -MP -FC -GS- -EHsc- -GR- -O2 -MT -Z7 -DNO_ARRAY_BOUNDS_CHECK
+set compiler_warnings= ^
+    -W4 -WX ^
+    -wd4100 -wd4101 -wd4127 -wd4189 ^
+    -wd4201 -wd4204 ^
+    -wd4456 -wd4457 -wd4480 ^
+    -wd4512
+
+set compiler_includes=
+set libs= ^
+    kernel32.lib
+
+set linker_flags= -incremental:no -opt:ref -subsystem:console -debug
+
+set compiler_settings=%compiler_includes% %compiler_flags% %compiler_warnings%
+set linker_settings=%libs% %linker_flags%
+
+cl %compiler_settings% "src\main.cpp" ^
+    /link %linker_settings% -OUT:%exe_name% ^
+

code/demo.odin

@@ -1,127 +0,0 @@
-#import "fmt.odin";
-
-main :: proc() {
-/*
-	Version 0.1.0
-
-	Added:
-	 * Dynamic Arrays `[...]Type`
-	 * Dynamic Maps   `map[Key]Value`
-	 * Dynamic array and map literals
-	 * Custom struct alignemnt `struct #align 8 { bar: i8 }`
-	 * Allow `_` in numbers
-	 * Variadic `append`
-	 * fmt.sprint*
-	 * Entities prefixes with an underscore do not get exported on imports
-	 * Overloaded `free` for pointers, slices, strings, dynamic arrays, and dynamic maps
-	 * enum types have an implict `names` field, a []string of all the names in that enum
-
-	Removed:
-	 * Maybe/option types
-	 * immutable variables
-	 * Remove `type` keyword and other "reserved" keywords
-	 * `compile_assert` and `assert`return the value of the condition for semantic reasons
-
-	Changed:
-	 * thread_local -> #thread_local
-	 * #include -> #load
-	 * Files only get checked if they are actually used
-	 * match x in y {} // For type match statements
-	 * Version numbering now starts from 0.1.0 and uses the convention:
-	 	- major.minor.patch
-
-	Fixes:
-	 * Many fmt.* fixes
-
-	To come very Soon™:
-	 * Linux and OS X builds (unofficial ones do exist already)
-*/
-
-	{
-		Fruit :: enum {
-			APPLE,
-			BANANA,
-			COCONUT,
-		}
-		fmt.println(Fruit.names);
-	}
-
-	{
-		m: map[f32]int;
-		reserve(m, 16);
-		defer free(m);
-
-		m[1.0] = 1278;
-		m[2.0] = 7643;
-		m[3.0] = 564;
-		_, ok := m[3.0];
-		c := m[3.0];
-		assert(ok && c == 564);
-
-		fmt.print("map[");
-		i := 0;
-		for val, key in m {
-			if i > 0 {
-				fmt.print(", ");
-			}
-			fmt.printf("%v=%v", key, val);
-			i += 1;
-		}
-		fmt.println("]");
-	}
-	{
-		m := map[string]u32{
-			"a" = 56,
-			"b" = 13453,
-			"c" = 7654,
-		};
-		defer free(m);
-
-		c := m["c"];
-		_, ok := m["c"];
-		assert(ok && c == 7654);
-
-		fmt.println(m);
-	}
-
-	{
-		x: [...]f64;
-		reserve(x, 16);
-		defer free(x);
-		append(x, 2_000_000.500_000, 3, 5, 7);
-
-		for p, i in x {
-			if i > 0 { fmt.print(", "); }
-			fmt.print(p);
-		}
-		fmt.println();
-	}
-
-	{
-		x := [...]f64{2_000_000.500_000, 3, 5, 7};
-		defer free(x);
-		fmt.println(x);
-	}
-
-
-	{
-		Vec3 :: [vector 3]f32;
-
-		x := Vec3{1, 2, 3};
-		y := Vec3{4, 5, 6};
-		fmt.println(x < y);
-		fmt.println(x + y);
-		fmt.println(x - y);
-		fmt.println(x * y);
-		fmt.println(x / y);
-
-		for i in x {
-			fmt.println(i);
-		}
-
-		compile_assert(size_of([vector 7]bool) == size_of([7]bool));
-		compile_assert(size_of([vector 7]i32) == size_of([7]i32));
-		// align_of([vector 7]i32) != align_of([7]i32) // this may be the case
-	}
-}
-

code/game.odin

@@ -1,215 +0,0 @@
-#import "win32.odin" when ODIN_OS == "windows";
-#import "fmt.odin";
-#import "math.odin";
-#import "os.odin";
-#import gl "opengl.odin";
-
-TWO_HEARTS :: '💕';
-
-win32_perf_count_freq := win32.GetQueryPerformanceFrequency();
-time_now :: proc() -> f64 {
-	assert(win32_perf_count_freq != 0);
-
-	counter: i64;
-	win32.QueryPerformanceCounter(^counter);
-	result := counter as f64 / win32_perf_count_freq as f64;
-	return result;
-}
-win32_print_last_error :: proc() {
-	err_code := win32.GetLastError() as int;
-	if err_code != 0 {
-		fmt.println("GetLastError: %", err_code);
-	}
-}
-
-// Yuk!
-to_c_string :: proc(s: string) -> []u8 {
-	c_str := new_slice(u8, s.count+1);
-	copy(c_str, s as []byte);
-	c_str[s.count] = 0;
-	return c_str;
-}
-
-
-Window :: struct {
-	width, height:      int;
-	wc:                 win32.WNDCLASSEXA;
-	dc:                 win32.HDC;
-	hwnd:               win32.HWND;
-	opengl_context, rc: win32.HGLRC;
-	c_title:            []u8;
-}
-
-make_window :: proc(title: string, msg, height: int, window_proc: win32.WNDPROC) -> (Window, bool) {
-	using win32;
-
-	w: Window;
-	w.width, w.height = msg, height;
-
-	class_name := "Win32-Odin-Window\x00";
-	c_class_name := class_name.data;
-	if title[title.count-1] != 0 {
-		w.c_title = to_c_string(title);
-	} else {
-		w.c_title = title as []u8;
-	}
-
-	instance := GetModuleHandleA(nil);
-
-	w.wc = WNDCLASSEXA{
-		size       = size_of(WNDCLASSEXA) as u32,
-		style      = CS_VREDRAW | CS_HREDRAW,
-		instance   = instance as HINSTANCE,
-		class_name = c_class_name,
-		wnd_proc   = window_proc,
-	};
-
-	if RegisterClassExA(^w.wc) == 0 {
-		win32_print_last_error();
-		return w, false;
-	}
-
-	w.hwnd = CreateWindowExA(0,
-	                         c_class_name, w.c_title.data,
-	                         WS_VISIBLE | WS_OVERLAPPED | WS_CAPTION | WS_SYSMENU | WS_MINIMIZEBOX,
-	                         CW_USEDEFAULT, CW_USEDEFAULT,
-	                         w.width as i32, w.height as i32,
-	                         nil, nil, instance, nil);
-
-	if w.hwnd == nil {
-		win32_print_last_error();
-		return w, false;
-	}
-
-	w.dc = GetDC(w.hwnd);
-
-	{
-		pfd := PIXELFORMATDESCRIPTOR{
-			size         = size_of(PIXELFORMATDESCRIPTOR) as u32,
-			version      = 1,
-			flags        = PFD_DRAW_TO_WINDOW | PFD_SUPPORT_OPENGL | PFD_DOUBLEBUFFER,
-			pixel_type   = PFD_TYPE_RGBA,
-			color_bits   = 32,
-			alpha_bits   = 8,
-			depth_bits   = 24,
-			stencil_bits = 8,
-			layer_type   = PFD_MAIN_PLANE,
-		};
-
-		SetPixelFormat(w.dc, ChoosePixelFormat(w.dc, ^pfd), nil);
-		w.opengl_context = wglCreateContext(w.dc);
-		wglMakeCurrent(w.dc, w.opengl_context);
-
-		attribs := [8]i32{
-			WGL_CONTEXT_MAJOR_VERSION_ARB, 2,
-			WGL_CONTEXT_MINOR_VERSION_ARB, 1,
-			WGL_CONTEXT_PROFILE_MASK_ARB, WGL_CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB,
-			0, // NOTE(bill): tells the proc that this is the end of attribs
-		};
-
-		wglCreateContextAttribsARB := wglGetProcAddress(("wglCreateContextAttribsARB\x00" as string).data) as wglCreateContextAttribsARBType;
-		w.rc = wglCreateContextAttribsARB(w.dc, 0, ^attribs[0]);
-		wglMakeCurrent(w.dc, w.rc);
-		SwapBuffers(w.dc);
-	}
-
-	return w, true;
-}
-
-destroy_window :: proc(w: ^Window) {
-	free(w.c_title.data);
-}
-
-display_window :: proc(w: ^Window) {
-	win32.SwapBuffers(w.dc);
-}
-
-
-run :: proc() {
-	using win32;
-	using math;
-
-	win32_proc :: proc(hwnd: HWND, msg: u32, wparam: WPARAM, lparam: LPARAM) -> LRESULT #no_inline {
-		if msg == WM_DESTROY || msg == WM_CLOSE || msg == WM_QUIT {
-			os.exit(0);
-			return 0;
-		}
-		return DefWindowProcA(hwnd, msg, wparam, lparam);
-	}
-
-	window, window_success := make_window("Odin Language Demo", 854, 480, win32_proc);
-	if !window_success {
-		return;
-	}
-	defer destroy_window(^window);
-
-	gl.init();
-
-
-	prev_time := time_now();
-	running := true;
-
-	pos := Vec2{100, 100};
-
-	for running {
-		curr_time := time_now();
-		dt := (curr_time - prev_time) as f32;
-		prev_time = curr_time;
-
-		msg: MSG;
-		for PeekMessageA(^msg, nil, 0, 0, PM_REMOVE) > 0 {
-			if msg.message == WM_QUIT {
-				running = false;
-			}
-			TranslateMessage(^msg);
-			DispatchMessageA(^msg);
-		}
-
-		if is_key_down(Key_Code.ESCAPE) {
-			running = false;
-		}
-
-		{
-			SPEED :: 500;
-			v: Vec2;
-
-			if is_key_down(Key_Code.RIGHT) { v[0] += 1; }
-			if is_key_down(Key_Code.LEFT)  { v[0] -= 1; }
-			if is_key_down(Key_Code.UP)    { v[1] += 1; }
-			if is_key_down(Key_Code.DOWN)  { v[1] -= 1; }
-
-			v = vec2_norm0(v);
-
-			pos += v * Vec2{SPEED * dt};
-		}
-
-
-		gl.ClearColor(0.5, 0.7, 1.0, 1.0);
-		gl.Clear(gl.COLOR_BUFFER_BIT);
-
-		gl.LoadIdentity();
-		gl.Ortho(0, window.width as f64,
-		         0, window.height as f64, 0, 1);
-
-		draw_rect :: proc(x, y, w, h: f32) {
-			gl.Begin(gl.TRIANGLES);
-			defer gl.End();
-
-			gl.Color3f(1, 0, 0); gl.Vertex3f(x,   y,   0);
-			gl.Color3f(0, 1, 0); gl.Vertex3f(x+w, y,   0);
-			gl.Color3f(0, 0, 1); gl.Vertex3f(x+w, y+h, 0);
-
-			gl.Color3f(0, 0, 1); gl.Vertex3f(x+w, y+h, 0);
-			gl.Color3f(1, 1, 0); gl.Vertex3f(x,   y+h, 0);
-			gl.Color3f(1, 0, 0); gl.Vertex3f(x,   y,   0);
-		}
-
-		draw_rect(pos.x, pos.y, 50, 50);
-
-		display_window(^window);
-		ms_to_sleep := (16 - 1000*dt) as i32;
-		if ms_to_sleep > 0 {
-			win32.Sleep(ms_to_sleep);
-		}
-	}
-}

code/http_test.odin

@@ -1,180 +0,0 @@
-#import "fmt.odin";
-
-#foreign_system_library ws2 "Ws2_32.lib" when ODIN_OS == "windows";
-
-
-SOCKET :: #type uint;
-INVALID_SOCKET :: ~(cast(SOCKET)0);
-
-AF :: enum i32 {
-	UNSPEC    = 0,       // unspecified
-	UNIX      = 1,       // local to host (pipes, portals)
-	INET      = 2,       // internetwork: UDP, TCP, etc.
-	IMPLINK   = 3,       // arpanet imp addresses
-	PUP       = 4,       // pup protocols: e.g. BSP
-	CHAOS     = 5,       // mit CHAOS protocols
-	NS        = 6,       // XEROX NS protocols
-	ISO       = 7,       // ISO protocols
-	OSI       = ISO,     // OSI is ISO
-	ECMA      = 8,       // european computer manufacturers
-	DATAKIT   = 9,       // datakit protocols
-	CCITT     = 10,      // CCITT protocols, X.25 etc
-	SNA       = 11,      // IBM SNA
-	DECnet    = 12,      // DECnet
-	DLI       = 13,      // Direct data link interface
-	LAT       = 14,      // LAT
-	HYLINK    = 15,      // NSC Hyperchannel
-	APPLETALK = 16,      // AppleTalk
-	ROUTE     = 17,      // Internal Routing Protocol
-	LINK      = 18,      // Link layer interface
-	XTP       = 19,      // eXpress Transfer Protocol (no AF)
-	COIP      = 20,      // connection-oriented IP, aka ST II
-	CNT       = 21,      // Computer Network Technology
-	RTIP      = 22,      // Help Identify RTIP packets
-	IPX       = 23,      // Novell Internet Protocol
-	SIP       = 24,      // Simple Internet Protocol
-	PIP       = 25,      // Help Identify PIP packets
-	MAX       = 26,
-};
-
-SOCK_STREAM  :: 1;
-SOCKET_ERROR :: -1;
-IPPROTO_TCP  :: 6;
-AI_PASSIVE   :: 0x0020;
-SOMAXCONN    :: 128;
-
-SD_RECEIVE :: 0;
-SD_SEND    :: 1;
-SD_BOTH    :: 2;
-
-WSADESCRIPTION_LEN :: 256;
-WSASYS_STATUS_LEN  :: 128;
-WSADATA :: struct #ordered {
-	version:       i16,
-	high_version:  i16,
-
-
-// NOTE(bill): This is x64 ordering
-	max_sockets:   u16,
-	max_udp_dg:    u16,
-	vendor_info:   ^byte,
-	description:   [WSADESCRIPTION_LEN+1]byte,
-	system_status: [WSASYS_STATUS_LEN+1]byte,
-}
-
-addrinfo :: struct #ordered {
-	flags:     i32,
-	family:    i32,
-	socktype:  i32,
-	protocol:  i32,
-	addrlen:   uint,
-	canonname: ^u8,
-	addr:      ^sockaddr,
-	next:      ^addrinfo,
-}
-
-sockaddr :: struct #ordered {
-	family: u16,
-	data:   [14]byte,
-}
-
-
-WSAStartup      :: proc(version_requested: i16, data: ^WSADATA) -> i32                             #foreign ws2;
-WSACleanup      :: proc() -> i32                                                                   #foreign ws2;
-getaddrinfo     :: proc(node_name, service_name: ^u8, hints: ^addrinfo, result: ^^addrinfo) -> i32 #foreign ws2;
-freeaddrinfo    :: proc(ai: ^addrinfo)                                                             #foreign ws2;
-socket          :: proc(af, type_, protocol: i32) -> SOCKET                                        #foreign ws2;
-closesocket     :: proc(s: SOCKET) -> i32                                                          #foreign ws2;
-bind            :: proc(s: SOCKET, name: ^sockaddr, name_len: i32) -> i32                          #foreign ws2;
-listen          :: proc(s: SOCKET, back_log: i32) -> i32                                           #foreign ws2;
-accept          :: proc(s: SOCKET, addr: ^sockaddr, addr_len: i32) -> SOCKET                       #foreign ws2;
-recv            :: proc(s: SOCKET, buf: ^byte, len: i32, flags: i32) -> i32                        #foreign ws2;
-send            :: proc(s: SOCKET, buf: ^byte, len: i32, flags: i32) -> i32                        #foreign ws2;
-shutdown        :: proc(s: SOCKET, how: i32) -> i32                                                #foreign ws2;
-WSAGetLastError :: proc() -> i32                                                                   #foreign ws2;
-
-to_c_string :: proc(s: string) -> ^byte {
-	c_str := new_slice(byte, s.count+1);
-	assert(c_str.data != nil);
-	copy(c_str, cast([]byte)s);
-	c_str[s.count] = 0;
-	return c_str.data;
-}
-
-run :: proc() {
-	wsa: WSADATA;
-	res:  ^addrinfo = nil;
-	hints: addrinfo;
-	s, client: SOCKET;
-
-	if WSAStartup(2 | (2 << 8), ^wsa) != 0 {
-		fmt.println("WSAStartup failed: ", WSAGetLastError());
-		return;
-	}
-	defer WSACleanup();
-
-	hints.family   = cast(i32)AF.INET;
-	hints.socktype = SOCK_STREAM;
-	hints.protocol = IPPROTO_TCP;
-	hints.flags    = AI_PASSIVE;
-
-	if getaddrinfo(nil, to_c_string("8080"), ^hints, ^res) != 0 {
-		fmt.println("getaddrinfo failed: ", WSAGetLastError());
-		return;
-	}
-	defer freeaddrinfo(res);
-
-	s = socket(res.family, res.socktype, res.protocol);
-	if s == INVALID_SOCKET {
-		fmt.println("socket failed: ", WSAGetLastError());
-		return;
-	}
-	defer closesocket(s);
-
-	bind(s, res.addr, cast(i32)res.addrlen);
-	listen(s, SOMAXCONN);
-
-	client = accept(s, nil, 0);
-	if client == INVALID_SOCKET {
-		fmt.println("socket failed: ", WSAGetLastError());
-		return;
-	}
-	defer closesocket(client);
-
-	html :=
-`HTTP/1.1 200 OK
-Connection: close
-Content-type: text/html
-
-<html>
-<head>
-	<title>Demo Title</title>
-</head>
-<body>
-	<h1 style="color: orange;">Odin Server Demo</h1>
-</body>
-</html>
-`;
-
-	buf: [1024]byte;
-	for {
-		bytes := recv(client, ^buf[0], cast(i32)buf.count, 0);
-		if bytes > 0 {
-			// fmt.println(buf[:bytes] as string)
-			bytes_sent := send(client, html.data, cast(i32)(html.count-1), 0);
-			if bytes_sent == SOCKET_ERROR {
-				fmt.println("send failed: ", WSAGetLastError());
-				return;
-			}
-			break;
-		} else if bytes == 0 {
-			fmt.println("Connection closing...");
-			break;
-		} else {
-			fmt.println("recv failed: ", WSAGetLastError());
-			return;
-		}
-	}
-
-	shutdown(client, SD_SEND);
-}

code/old_demos/demo002.odin

@@ -1,900 +0,0 @@
-// Demo 002
-#include "basic.odin"
-#include "math.odin"
-// #include "game.odin"
-
-#thread_local tls_int: int
-
-main :: proc() {
-	// Forenotes
-
-	// Semicolons are now optional
-	// Rule for when a semicolon is expected after a statement
-	// - If the next token is not on the same line
-	// - if the next token is a closing brace }
-	// - Otherwise, a semicolon is needed
-	//
-	// Expections:
-	// for, if, match
-	// if x := thing(); x < 123 {}
-	// for i := 0; i < 123; i++ {}
-
-	// Q: Should I use the new rule or go back to the old one without optional semicolons?
-
-
-	// #thread_local - see runtime.odin and above at `tls_int`
-	// #foreign_system_library - see win32.odin
-
-	// struct_compound_literals()
-	// enumerations()
-	// variadic_procedures()
-	// new_builtins()
-	// match_statement()
-	// namespacing()
-	// subtyping()
-	// tagged_unions()
-}
-
-struct_compound_literals :: proc() {
-	Thing :: type struct {
-		id: int
-		x: f32
-		name: string
-	}
-	{
-		t1: Thing
-		t1.id = 1
-
-		t3 := Thing{}
-		t4 := Thing{1, 2, "Fred"}
-		// t5 := Thing{1, 2}
-
-		t6 := Thing{
-			name = "Tom",
-			x    = 23,
-		}
-	}
-}
-
-enumerations :: proc() {
-	{
-		Fruit :: type enum {
-			APPLE,  // 0
-			BANANA, // 1
-			PEAR,   // 2
-		}
-
-		f := Fruit.APPLE
-		// g12: int = Fruit.BANANA
-		g: int = Fruit.BANANA as int
-		// However, you can use enums are index values as _any_ integer allowed
-	}
-	{
-		Fruit1 :: type enum int {
-			APPLE,
-			BANANA,
-			PEAR,
-		}
-
-		Fruit2 :: type enum u8 {
-			APPLE,
-			BANANA,
-			PEAR,
-		}
-
-		Fruit3 :: type enum u8 {
-			APPLE = 1,
-			BANANA, // 2
-			PEAR  = 5,
-			TOMATO, // 6
-		}
-	}
-
-	// Q: remove the need for `type` if it's a record (struct/enum/raw_union/union)?
-}
-
-variadic_procedures :: proc() {
-	print_ints :: proc(args: ..int) {
-		for i := 0; i < len(args); i++ {
-			if i > 0 {
-				print_string(", ")
-			}
-			print_int(args[i])
-		}
-	}
-
-	print_ints(); // nl()
-	print_ints(1); nl()
-	print_ints(1, 2, 3); nl()
-
-	print_prefix_f32s :: proc(prefix: string, args: ..f32) {
-		print_string(prefix)
-		print_string(": ")
-		for i := 0; i < len(args); i++ {
-			if i > 0 {
-				print_string(", ")
-			}
-			print_f32(args[i])
-		}
-	}
-
-	print_prefix_f32s("a"); nl()
-	print_prefix_f32s("b", 1); nl()
-7	print_prefix_f32s("c", 1, 2, 3); nl()
-
-	// Internally, the variadic procedures get allocated to an array on the stack,
-	// and this array is passed a slice
-
-	// This is first step for a `print` procedure but I do not have an `any` type
-	// yet as this requires a few other things first - i.e. introspection
-
-	// NOTE(bill): I haven't yet added the feature of expanding a slice or array into
-	// a variadic a parameter but it's pretty trivial to add
-}
-
-new_builtins :: proc() {
-	{
-		a := new(int)
-		b := new_slice(int, 12)
-		c := new_slice(int, 12, 16)
-
-		defer delete(a)
-		defer delete(b)
-		defer delete(c)
-
-		// NOTE(bill): These use the current context's allocator not the default allocator
-		// see runtime.odin
-
-		// Q: Should this be `free` rather than `delete` and should I overload it for slices too?
-
-		{
-			prev_context := context
-			defer context = prev_context
-			// Q: Should I add a `push_context` feature to the language?
-
-			context.allocator = __default_allocator()
-
-			a := new(int)
-			defer delete(a)
-
-			// Do whatever
-
-		}
-	}
-
-	{
-		a: int = 123
-		b: type_of_val(a) = 321
-
-		// NOTE(bill): This matches the current naming scheme
-		// size_of
-		// align_of
-		// offset_of
-		//
-		// size_of_val
-		// align_of_val
-		// offset_of_val
-		// type_of_val
-	}
-
-	{
-		// Compile time assert
-		COND :: true
-		compile_assert(COND)
-		// compile_assert(!COND)
-
-		// Runtime assert
-		x := true
-		assert(x)
-		// assert(!x)
-	}
-
-	{
-		x: ^u32 = null;
-		y := ptr_offset(x, 100)
-		z := ptr_sub(y, x)
-		w := slice_ptr(x, 12)
-		t := slice_ptr(x, 12, 16)
-
-		// NOTE(bill): These are here because I've removed:
-		// pointer arithmetic
-		// pointer indexing
-		// pointer slicing
-
-		// Reason
-
-		a: [16]int
-		a[1] = 1;
-		b := ^a
-		// Auto pointer deref
-		// consistent with record members
-		assert(b[1] == 1)
-
-		// Q: Should I add them back in at the cost of inconsitency?
-	}
-
-	{
-		a, b := -1, 2
-		print_int(min(a, b)); nl()
-		print_int(max(a, b)); nl()
-		print_int(abs(a)); nl()
-
-		// These work at compile time too
-		A :: -1
-		B :: 2
-		C :: min(A, B)
-		D :: max(A, B)
-		E :: abs(A)
-
-		print_int(C); nl()
-		print_int(D); nl()
-		print_int(E); nl()
-	}
-}
-
-
-match_statement :: proc() {
-	// NOTE(bill): `match` statements are similar to `switch` statements
-	// in other languages but there are few differences
-
-	{
-		match x := 5; x {
-		case 1: // cases must be constant expression
-			print_string("1!\n")
-			// break by default
-
-		case 2:
-			s := "2!\n"; // Each case has its own scope
-			print_string(s)
-			break // explicit break
-
-		case 3, 4: // multiple cases
-			print_string("3 or 4!\n")
-
-		case 5:
-			print_string("5!\n")
-			fallthrough // explicit fallthrough
-
-		default:
-			print_string("default!\n")
-		}
-
-
-
-		match x := 1.5; x {
-		case 1.5:
-			print_string("1.5!\n")
-			// break by default
-		case MATH_TAU:
-			print_string("τ!\n")
-		default:
-			print_string("default!\n")
-		}
-
-
-
-		match x := "Hello"; x {
-		case "Hello":
-			print_string("greeting\n")
-			// break by default
-		case "Goodbye":
-			print_string("farewell\n")
-		default:
-			print_string("???\n")
-		}
-
-
-
-
-
-
-		a := 53
-		match {
-		case a == 1:
-			print_string("one\n")
-		case a == 2:
-			print_string("a couple\n")
-		case a < 7, a == 7:
-			print_string("a few\n")
-		case a < 12: // intentional bug
-			print_string("several\n")
-		case a >= 12 && a < 100:
-			print_string("dozens\n")
-		case a >= 100 && a < 1000:
-			print_string("hundreds\n")
-		default:
-			print_string("a fuck ton\n")
-		}
-
-		// Identical to this
-
-		b := 53
-		if b == 1 {
-			print_string("one\n")
-		} else if b == 2 {
-			print_string("a couple\n")
-		} else if b < 7 || b == 7 {
-			print_string("a few\n")
-		} else if b < 12 { // intentional bug
-			print_string("several\n")
-		} else if b >= 12 && b < 100 {
-			print_string("dozens\n")
-		} else if b >= 100 && b < 1000 {
-			print_string("hundreds\n")
-		} else {
-			print_string("a fuck ton\n")
-		}
-
-		// However, match statements allow for `break` and `fallthrough` unlike
-		// an if statement
-	}
-}
-
-Vector3 :: type struct {
-	x, y, z: f32
-}
-
-print_floats :: proc(args: ..f32) {
-	for i := 0; i < len(args); i++ {
-		if i > 0 {
-			print_string(", ")
-		}
-		print_f32(args[i])
-	}
-	print_nl()
-}
-
-namespacing :: proc() {
-	{
-		Thing :: type struct {
-			x: f32
-			name: string
-		}
-
-		a: Thing
-		a.x = 3
-		{
-			Thing :: type struct {
-				y: int
-				test: bool
-			}
-
-			b: Thing // Uses this scope's Thing
-			b.test = true
-		}
-	}
-
-	{
-		Entity :: type struct {
-			Guid :: type int
-			Nested :: type struct {
-				MyInt :: type int
-				i: int
-			}
-
-			CONSTANT :: 123
-
-
-			guid:   Guid
-			name:   string
-			pos:    Vector3
-			vel:    Vector3
-			nested: Nested
-		}
-
-		guid: Entity.Guid = Entity.CONSTANT
-		i: Entity.Nested.MyInt
-
-
-
-		{
-			using Entity
-			guid: Guid = CONSTANT
-			using Nested
-			i: MyInt
-		}
-
-
-		{
-			using Entity.Nested
-			guid: Entity.Guid = Entity.CONSTANT
-			i: MyInt
-		}
-
-
-		{
-			e: Entity
-			using e
-			guid = 27832
-			name = "Bob"
-
-			print_int(e.guid as int); nl()
-			print_string(e.name); nl()
-		}
-
-		{
-			using e: Entity
-			guid = 78456
-			name = "Thing"
-
-			print_int(e.guid as int); nl()
-			print_string(e.name); nl()
-		}
-	}
-
-	{
-		Entity :: type struct {
-			Guid :: type int
-			Nested :: type struct {
-				MyInt :: type int
-				i: int
-			}
-
-			CONSTANT :: 123
-
-
-			guid:      Guid
-			name:      string
-			using pos: Vector3
-			vel:       Vector3
-			using nested: ^Nested
-		}
-
-		e := Entity{nested = new(Entity.Nested)}
-		e.x = 123
-		e.i = Entity.CONSTANT
-	}
-
-
-
-	{
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		print_pos_1 :: proc(entity: ^Entity) {
-			print_string("print_pos_1: ")
-			print_floats(entity.position.x, entity.position.y, entity.position.z)
-		}
-
-		print_pos_2 :: proc(entity: ^Entity) {
-			using entity
-			print_string("print_pos_2: ")
-			print_floats(position.x, position.y, position.z)
-		}
-
-		print_pos_3 :: proc(using entity: ^Entity) {
-			print_string("print_pos_3: ")
-			print_floats(position.x, position.y, position.z)
-		}
-
-		print_pos_4 :: proc(using entity: ^Entity) {
-			using position
-			print_string("print_pos_4: ")
-			print_floats(x, y, z)
-		}
-
-		e := Entity{position = Vector3{1, 2, 3}}
-		print_pos_1(^e)
-		print_pos_2(^e)
-		print_pos_3(^e)
-		print_pos_4(^e)
-
-		// This is similar to C++'s `this` pointer that is implicit and only available in methods
-	}
-}
-
-subtyping :: proc() {
-	{
-		// C way for subtyping/subclassing
-
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		Frog :: type struct {
-			entity: Entity
-			jump_height: f32
-		}
-
-		f: Frog
-		f.entity.position = Vector3{1, 2, 3}
-
-		using f.entity
-		position = Vector3{1, 2, 3}
-
-	}
-
-	{
-		// C++ way for subtyping/subclassing
-
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		Frog :: type struct {
-			using entity: Entity
-			jump_height: f32
-		}
-
-		f: Frog
-		f.position = Vector3{1, 2, 3}
-
-
-		print_pos :: proc(using entity: Entity) {
-			print_string("print_pos: ")
-			print_floats(position.x, position.y, position.z)
-		}
-
-		print_pos(f.entity)
-		print_pos(f)
-
-		// Subtype Polymorphism
-	}
-
-	{
-		// More than C++ way for subtyping/subclassing
-
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		Frog :: type struct {
-			jump_height: f32
-			using entity: ^Entity // Doesn't have to be first member!
-		}
-
-		f: Frog
-		f.entity = new(Entity)
-		f.position = Vector3{1, 2, 3}
-
-
-		print_pos :: proc(using entity: ^Entity) {
-			print_string("print_pos: ")
-			print_floats(position.x, position.y, position.z)
-		}
-
-		print_pos(f.entity)
-		print_pos(^f)
-		print_pos(f)
-	}
-
-	{
-		// More efficient subtyping
-
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		Frog :: type struct {
-			jump_height: f32
-			using entity: ^Entity
-		}
-
-		MAX_ENTITES :: 64
-		entities: [MAX_ENTITES]Entity
-		entity_count := 0
-
-		next_entity :: proc(entities: []Entity, entity_count: ^int) -> ^Entity {
-			e := ^entities[entity_count^]
-			entity_count^++
-			return e
-		}
-
-		f: Frog
-		f.entity = next_entity(entities[:], ^entity_count)
-		f.position = Vector3{3, 4, 6}
-
-		using f.position
-		print_floats(x, y, z)
-	}
-
-	{
-		// Down casting
-
-		Entity :: type struct {
-			position: Vector3
-		}
-
-		Frog :: type struct {
-			jump_height: f32
-			using entity: Entity
-		}
-
-		f: Frog
-		f.jump_height = 564
-		e := ^f.entity
-
-		frog := e down_cast ^Frog
-		print_string("down_cast: ")
-		print_f32(frog.jump_height); nl()
-
-		// NOTE(bill): `down_cast` is unsafe and there are not check are compile time or run time
-		// Q: Should I completely remove `down_cast` as I added it in about 30 minutes
-	}
-
-	{
-		// Multiple "inheritance"/subclassing
-
-		Entity :: type struct {
-			position: Vector3
-		}
-		Climber :: type struct {
-			speed: f32
-		}
-
-		Frog :: type struct {
-			using entity:  Entity
-			using climber: Climber
-		}
-	}
-}
-
-tagged_unions :: proc() {
-	{
-		EntityKind :: type enum {
-			INVALID,
-			FROG,
-			GIRAFFE,
-			HELICOPTER,
-		}
-
-		Entity :: type struct {
-			kind: EntityKind
-			using data: raw_union {
-				frog: struct {
-					jump_height: f32
-					colour: u32
-				}
-				giraffe: struct {
-					neck_length: f32
-					spot_count: int
-				}
-				helicopter: struct {
-					blade_count: int
-					weight: f32
-					pilot_name: string
-				}
-			}
-		}
-
-		e: Entity
-		e.kind = EntityKind.FROG
-		e.frog.jump_height = 12
-
-		f: type_of_val(e.frog);
-
-		// But this is very unsafe and extremely cumbersome to write
-		// In C++, I use macros to alleviate this but it's not a solution
-	}
-
-	{
-		Entity :: type union {
-			Frog: struct {
-				jump_height: f32
-				colour: u32
-			}
-			Giraffe: struct {
-				neck_length: f32
-				spot_count: int
-			}
-			Helicopter: struct {
-				blade_count: int
-				weight: f32
-				pilot_name: string
-			}
-		}
-
-		using Entity
-		f1: Frog = Frog{12, 0xff9900}
-		f2: Entity = Frog{12, 0xff9900} // Implicit cast
-		f3 := Frog{12, 0xff9900} as Entity // Explicit cast
-
-		// f3.Frog.jump_height = 12 // There are "members" of a union
-
-
-
-		e, f, g, h: Entity
-		f = Frog{12, 0xff9900}
-		g = Giraffe{2.1, 23}
-		h = Helicopter{4, 1000, "Frank"}
-
-
-
-
-		// Requires a pointer to the union
-		// `x` will be a pointer to type of the case
-
-		match type x : ^f {
-		case Frog:
-			print_string("Frog!\n")
-			print_f32(x.jump_height); nl()
-			x.jump_height = 3
-			print_f32(x.jump_height); nl()
-		case Giraffe:
-			print_string("Giraffe!\n")
-		case Helicopter:
-			print_string("ROFLCOPTER!\n")
-		default:
-			print_string("invalid entity\n")
-		}
-
-
-		// Q: Allow for a non pointer version with takes a copy instead?
-		// Or it takes the pointer the data and not a copy
-
-
-		fp := ^f as ^Frog // Unsafe
-		print_f32(fp.jump_height); nl()
-
-
-		// Internals of a tagged union
-		/*
-			struct {
-				data: [size_of_biggest_tag]u8
-				tag_index: int
-			}
-		*/
-		// This is to allow for pointer casting if needed
-
-
-		// Advantage over subtyping version
-		MAX_ENTITES :: 64
-		entities: [MAX_ENTITES]Entity
-
-		entities[0] = Frog{}
-		entities[1] = Helicopter{}
-		// etc.
-	}
-
-
-	{
-		// Transliteration of code from this actual compiler
-		// Some stuff is missing
-		Type       :: type struct {}
-		Scope      :: type struct {}
-		Token      :: type struct {}
-		AstNode    :: type struct {}
-		ExactValue :: type struct {}
-
-		EntityKind :: type enum {
-			Invalid,
-			Constant,
-			Variable,
-			UsingVariable,
-			TypeName,
-			Procedure,
-			Builtin,
-			Count,
-		}
-
-		Entity :: type struct {
-			Guid :: type i64
-
-			kind: EntityKind
-			guid: Guid
-
-			scope: ^Scope
-			token: Token
-			type_: ^Type
-
-			using data: raw_union {
-				Constant: struct {
-					value: ExactValue
-				}
-				Variable: struct {
-					visited:   bool // Cycle detection
-					used:      bool // Variable is used
-					is_field:  bool // Is struct field
-					anonymous: bool // Variable is an anonymous
-				}
-				UsingVariable: struct {
-				}
-				TypeName: struct {
-				}
-				Procedure: struct {
-					used: bool
-				}
-				Builtin: struct {
-					id: int
-				}
-			}
-		}
-
-		// Plus all the constructing procedures that go along with them!!!!
-		// It's a nightmare
-	}
-
-	{
-		Type       :: type struct {}
-		Scope      :: type struct {}
-		Token      :: type struct {}
-		AstNode    :: type struct {}
-		ExactValue :: type struct {}
-
-
-		Entity :: type union {
-			Base :: type struct {
-				Guid :: type i64
-				guid: Guid
-
-				scope: ^Scope
-				token: Token
-				type_: ^Type
-			}
-
-
-			Constant: struct {
-				using base: Base
-				value: ExactValue
-			}
-			Variable: struct {
-				using base: Base
-				visited:   bool // Cycle detection
-				used:      bool // Variable is used
-				is_field:  bool // Is struct field
-				anonymous: bool // Variable is an anonymous
-			}
-			UsingVariable: struct {
-				using base: Base
-			}
-			TypeName: struct {
-				using base: Base
-			}
-			Procedure: struct {
-				using base: Base
-				used: bool
-			}
-			Builtin: struct {
-				using base: Base
-				id: int
-			}
-		}
-
-		using Entity
-
-		e: Entity
-
-		e = Variable{
-			base = Base{},
-			used = true,
-			anonymous = false,
-		}
-
-
-
-		// Q: Allow a "base" type to be added to a union?
-		// Or even `using` on union to get the same properties?
-	}
-
-
-	{
-		// `Raw` unions still have uses, especially for mathematic types
-
-		Vector2 :: type raw_union {
-			using xy_: struct { x, y: f32 }
-			e: [2]f32
-			v: {2}f32
-		}
-
-		Vector3 :: type raw_union {
-			using xyz_: struct { x, y, z: f32 }
-			xy: Vector2
-			e: [3]f32
-			v: {3}f32
-		}
-
-		v2: Vector2
-		v2.x = 1
-		v2.e[0] = 1
-		v2.v[0] = 1
-
-		v3: Vector3
-		v3.x = 1
-		v3.e[0] = 1
-		v3.v[0] = 1
-		v3.xy.x = 1
-	}
-}
-
-nl :: proc() { print_nl() }

code/punity.odin

@@ -1,489 +0,0 @@
-#import win32 "sys/windows.odin";
-#import "fmt.odin";
-#import "os.odin";
-#import "mem.odin";
-
-CANVAS_WIDTH  :: 128;
-CANVAS_HEIGHT :: 128;
-CANVAS_SCALE  :: 3;
-FRAME_TIME    :: 1.0/30.0;
-WINDOW_TITLE  :: "Punity\x00";
-
-_ := compile_assert(CANVAS_WIDTH % 16 == 0);
-
-WINDOW_WIDTH  :: CANVAS_WIDTH  * CANVAS_SCALE;
-WINDOW_HEIGHT :: CANVAS_HEIGHT * CANVAS_SCALE;
-
-
-STACK_CAPACITY   :: 1<<20;
-STORAGE_CAPACITY :: 1<<20;
-
-DRAW_LIST_RESERVE :: 128;
-
-MAX_KEYS :: 256;
-
-Core :: struct {
-	stack:   ^Bank,
-	storage: ^Bank,
-
-	running:       bool,
-	key_modifiers: u32,
-	key_states:    [MAX_KEYS]byte,
-	key_deltas:    [MAX_KEYS]byte,
-
-	perf_frame,
-	perf_frame_inner,
-	perf_step,
-	perf_audio,
-	perf_blit,
-	perf_blit_cvt,
-	perf_blit_gdi: Perf_Span,
-
-	frame: i64,
-
-	canvas:    Canvas,
-	draw_list: ^Draw_List,
-}
-
-Perf_Span :: struct {
-	stamp: f64,
-	delta: f32,
-}
-
-Bank :: struct {
-	memory: []byte,
-	cursor: int,
-}
-
-Bank_State :: struct {
-	state: Bank,
-	bank: ^Bank,
-}
-
-
-Color :: raw_union {
-	using channels: struct{a, b, g, r: byte},
-	rgba: u32,
-}
-
-Palette :: struct {
-	colors: [256]Color,
-	colors_count: byte,
-}
-
-
-Rect :: raw_union {
-	using minmax: struct {min_x, min_y, max_x, max_y: int},
-	using pos: struct {left, top, right, bottom: int},
-	e: [4]int,
-}
-
-Bitmap :: struct {
-	pixels: []byte,
-	width:  int,
-	height: int,
-}
-
-Font :: struct {
-	using bitmap: Bitmap,
-	char_width:   int,
-	char_height:  int,
-}
-
-Canvas :: struct {
-	using bitmap: ^Bitmap,
-	palette:      Palette,
-	translate_x:  int,
-	translate_y:  int,
-	clip:         Rect,
-	font:         ^Font,
-}
-
-DrawFlag :: enum {
-	NONE   = 0,
-	FLIP_H = 1<<0,
-	FLIP_V = 1<<1,
-	MASK   = 1<<2,
-}
-
-Draw_Item :: struct {}
-Draw_List :: struct {
-	items: []Draw_Item,
-}
-
-Key :: enum {
-	MOD_SHIFT   = 0x0001,
-	MOD_CONTROL = 0x0002,
-	MOD_ALT     = 0x0004,
-	MOD_SUPER   = 0x0008,
-
-	UNKNOWN            =-1,
-	INVALID            =-2,
-
-	LBUTTON            = 1,
-	RBUTTON            = 2,
-	CANCEL             = 3,
-	MBUTTON            = 4,
-
-	BACK               = 8,
-	TAB                = 9,
-	CLEAR              = 12,
-	RETURN             = 13,
-	SHIFT              = 16,
-	CONTROL            = 17,
-	MENU               = 18,
-	PAUSE              = 19,
-	CAPITAL            = 20,
-	KANA               = 0x15,
-	HANGEUL            = 0x15,
-	HANGUL             = 0x15,
-	JUNJA              = 0x17,
-	FINAL              = 0x18,
-	HANJA              = 0x19,
-	KANJI              = 0x19,
-	ESCAPE             = 0x1B,
-	CONVERT            = 0x1C,
-	NONCONVERT         = 0x1D,
-	ACCEPT             = 0x1E,
-	MODECHANGE         = 0x1F,
-	SPACE              = 32,
-	PRIOR              = 33,
-	NEXT               = 34,
-	END                = 35,
-	HOME               = 36,
-	LEFT               = 37,
-	UP                 = 38,
-	RIGHT              = 39,
-	DOWN               = 40,
-	SELECT             = 41,
-	PRINT              = 42,
-	EXEC               = 43,
-	SNAPSHOT           = 44,
-	INSERT             = 45,
-	DELETE             = 46,
-	HELP               = 47,
-	LWIN               = 0x5B,
-	RWIN               = 0x5C,
-	APPS               = 0x5D,
-	SLEEP              = 0x5F,
-	NUMPAD0            = 0x60,
-	NUMPAD1            = 0x61,
-	NUMPAD2            = 0x62,
-	NUMPAD3            = 0x63,
-	NUMPAD4            = 0x64,
-	NUMPAD5            = 0x65,
-	NUMPAD6            = 0x66,
-	NUMPAD7            = 0x67,
-	NUMPAD8            = 0x68,
-	NUMPAD9            = 0x69,
-	MULTIPLY           = 0x6A,
-	ADD                = 0x6B,
-	SEPARATOR          = 0x6C,
-	SUBTRACT           = 0x6D,
-	DECIMAL            = 0x6E,
-	DIVIDE             = 0x6F,
-	F1                 = 0x70,
-	F2                 = 0x71,
-	F3                 = 0x72,
-	F4                 = 0x73,
-	F5                 = 0x74,
-	F6                 = 0x75,
-	F7                 = 0x76,
-	F8                 = 0x77,
-	F9                 = 0x78,
-	F10                = 0x79,
-	F11                = 0x7A,
-	F12                = 0x7B,
-	F13                = 0x7C,
-	F14                = 0x7D,
-	F15                = 0x7E,
-	F16                = 0x7F,
-	F17                = 0x80,
-	F18                = 0x81,
-	F19                = 0x82,
-	F20                = 0x83,
-	F21                = 0x84,
-	F22                = 0x85,
-	F23                = 0x86,
-	F24                = 0x87,
-	NUMLOCK            = 0x90,
-	SCROLL             = 0x91,
-	LSHIFT             = 0xA0,
-	RSHIFT             = 0xA1,
-	LCONTROL           = 0xA2,
-	RCONTROL           = 0xA3,
-	LMENU              = 0xA4,
-	RMENU              = 0xA5,
-
-	APOSTROPHE         = 39,  /* ' */
-	COMMA              = 44,  /* , */
-	MINUS              = 45,  /* - */
-	PERIOD             = 46,  /* . */
-	SLASH              = 47,  /* / */
-	NUM0               = 48,
-	NUM1               = 49,
-	NUM2               = 50,
-	NUM3               = 51,
-	NUM4               = 52,
-	NUM5               = 53,
-	NUM6               = 54,
-	NUM7               = 55,
-	NUM8               = 56,
-	NUM9               = 57,
-	SEMICOLON          = 59,  /* ; */
-	EQUAL              = 61,  /* = */
-	A                  = 65,
-	B                  = 66,
-	C                  = 67,
-	D                  = 68,
-	E                  = 69,
-	F                  = 70,
-	G                  = 71,
-	H                  = 72,
-	I                  = 73,
-	J                  = 74,
-	K                  = 75,
-	L                  = 76,
-	M                  = 77,
-	N                  = 78,
-	O                  = 79,
-	P                  = 80,
-	Q                  = 81,
-	R                  = 82,
-	S                  = 83,
-	T                  = 84,
-	U                  = 85,
-	V                  = 86,
-	W                  = 87,
-	X                  = 88,
-	Y                  = 89,
-	Z                  = 90,
-	LEFT_BRACKET       = 91,  /* [ */
-	BACKSLASH          = 92,  /* \ */
-	RIGHT_BRACKET      = 93,  /* ] */
-	GRAVE_ACCENT       = 96,  /* ` */
-};
-
-
-key_down :: proc(k: Key) -> bool {
-	return _core.key_states[k] != 0;
-}
-
-key_pressed :: proc(k: Key) -> bool {
-	return (_core.key_deltas[k] != 0) && key_down(k);
-}
-
-
-
-
-win32_perf_count_freq := win32.GetQueryPerformanceFrequency();
-time_now :: proc() -> f64 {
-	assert(win32_perf_count_freq != 0);
-
-	counter: i64;
-	win32.QueryPerformanceCounter(^counter);
-	result := cast(f64)counter / cast(f64)win32_perf_count_freq;
-	return result;
-}
-
-_core: Core;
-
-run :: proc(user_init, user_step: proc(c: ^Core)) {
-	using win32;
-
-	_core.running = true;
-
-	win32_proc :: proc(hwnd: win32.HWND, msg: u32, wparam: win32.WPARAM, lparam: win32.LPARAM) -> win32.LRESULT #no_inline #cc_c {
-		win32_app_key_mods :: proc() -> u32 {
-			mods: u32 = 0;
-
-			if is_key_down(Key_Code.SHIFT) {
-				mods |= cast(u32)Key.MOD_SHIFT;
-			}
-			if is_key_down(Key_Code.CONTROL) {
-				mods |= cast(u32)Key.MOD_CONTROL;
-			}
-			if is_key_down(Key_Code.MENU) {
-				mods |= cast(u32)Key.MOD_ALT;
-			}
-			if is_key_down(Key_Code.LWIN) || is_key_down(Key_Code.RWIN) {
-				mods |= cast(u32)Key.MOD_SUPER;
-			}
-
-			return mods;
-		}
-
-		match msg {
-		case WM_KEYDOWN:
-			_core.key_modifiers = win32_app_key_mods();
-			if wparam < MAX_KEYS {
-				_core.key_states[wparam] = 1;
-				_core.key_deltas[wparam] = 1;
-			}
-			return 0;
-
-		case WM_KEYUP:
-			_core.key_modifiers = win32_app_key_mods();
-			if wparam < MAX_KEYS {
-				_core.key_states[wparam] = 0;
-				_core.key_deltas[wparam] = 1;
-			}
-			return 0;
-
-		case WM_CLOSE:
-			PostQuitMessage(0);
-			_core.running = false;
-			return 0;
-		}
-
-		return DefWindowProcA(hwnd, msg, wparam, lparam);
-	}
-
-
-	window_class := WNDCLASSEXA{
-		class_name = (cast(string)"Punity\x00").data, // C-style string
-		size       = size_of(WNDCLASSEXA),
-		style      = CS_HREDRAW | CS_VREDRAW | CS_OWNDC,
-		instance   = cast(HINSTANCE)GetModuleHandleA(nil),
-		wnd_proc   = win32_proc,
-		// wnd_proc   = DefWindowProcA,
-		background = cast(HBRUSH)GetStockObject(BLACK_BRUSH),
-	};
-
-	if RegisterClassExA(^window_class) == 0 {
-		fmt.fprintln(os.stderr, "RegisterClassExA failed");
-		return;
-	}
-
-	screen_width  := GetSystemMetrics(SM_CXSCREEN);
-	screen_height := GetSystemMetrics(SM_CYSCREEN);
-
-	rc: RECT;
-	rc.left   = (screen_width - WINDOW_WIDTH)   / 2;
-	rc.top    = (screen_height - WINDOW_HEIGHT) / 2;
-	rc.right  = rc.left + WINDOW_WIDTH;
-	rc.bottom = rc.top + WINDOW_HEIGHT;
-
-	style: u32 = WS_CAPTION | WS_SYSMENU | WS_MINIMIZEBOX;
-	assert(AdjustWindowRect(^rc, style, 0) != 0);
-
-	wt := WINDOW_TITLE;
-
-	win32_window := CreateWindowExA(0,
-	                                window_class.class_name,
-	                                wt.data,
-	                                style,
-	                                rc.left, rc.top,
-	                                rc.right-rc.left, rc.bottom-rc.top,
-	                                nil, nil, window_class.instance,
-	                                nil);
-
-	if win32_window == nil {
-		fmt.fprintln(os.stderr, "CreateWindowExA failed");
-		return;
-	}
-
-
-	window_bmi: BITMAPINFO;
-	window_bmi.size        = size_of(BITMAPINFOHEADER);
-	window_bmi.width       = CANVAS_WIDTH;
-	window_bmi.height      = CANVAS_HEIGHT;
-	window_bmi.planes      = 1;
-	window_bmi.bit_count   = 32;
-	window_bmi.compression = BI_RGB;
-
-
-	user_init(^_core);
-
-	ShowWindow(win32_window, SW_SHOW);
-
-	window_buffer := new_slice(u32, CANVAS_WIDTH * CANVAS_HEIGHT);
-	defer free(window_buffer);
-
-
-	for i := 0; i < window_buffer.count; i += 1 {
-		window_buffer[i] = 0xff00ff;
-	}
-
-
-	dt: f64;
-	prev_time := time_now();
-	curr_time := time_now();
-	total_time : f64 = 0;
-	offset_x := 0;
-	offset_y := 0;
-
-	message: MSG;
-	for _core.running {
-		curr_time = time_now();
-		dt = curr_time - prev_time;
-		prev_time = curr_time;
-		total_time += dt;
-
-		offset_x += 1;
-		offset_y += 2;
-
-		{
-			data: [128]byte;
-			buf: fmt.Buffer;
-			buf.data = data[:];
-			fmt.bprintf(^buf, "Punity: %.4f ms\x00", dt*1000);
-			win32.SetWindowTextA(win32_window, ^buf[0]);
-		}
-
-
-		for y := 0; y < CANVAS_HEIGHT; y += 1 {
-			for x := 0; x < CANVAS_WIDTH; x += 1 {
-				g := (x % 32) * 8;
-				b := (y % 32) * 8;
-				window_buffer[x + y*CANVAS_WIDTH] = cast(u32)(g << 8 | b);
-			}
-		}
-
-		mem.zero(^_core.key_deltas[0], size_of_val(_core.key_deltas));
-
-		for PeekMessageA(^message, nil, 0, 0, PM_REMOVE) != 0 {
-			if message.message == WM_QUIT {
-				_core.running = false;
-			}
-			TranslateMessage(^message);
-			DispatchMessageA(^message);
-		}
-
-		user_step(^_core);
-
-		dc := GetDC(win32_window);
-		StretchDIBits(dc,
-		              0, 0, CANVAS_WIDTH * CANVAS_SCALE, CANVAS_HEIGHT * CANVAS_SCALE,
-		              0, 0, CANVAS_WIDTH, CANVAS_HEIGHT,
-		              window_buffer.data,
-		              ^window_bmi,
-		              DIB_RGB_COLORS,
-		              SRCCOPY);
-		ReleaseDC(win32_window, dc);
-
-
-		{
-			delta := time_now() - prev_time;
-			ms := cast(i32)((FRAME_TIME - delta) * 1000);
-			if ms > 0 {
-				win32.Sleep(ms);
-			}
-		}
-
-		_core.frame += 1;
-	}
-}
-
-
-main :: proc() {
-	user_init :: proc(c: ^Core) {
-
-	}
-
-	user_step :: proc(c: ^Core) {
-
-	}
-
-	run(user_init, user_step);
-}

code/sub/test.odin

@@ -1,5 +0,0 @@
-#import "fmt.odin" as fmt
-
-thing :: proc() {
-	fmt.println("Sub Hello!")
-}

code/test.odin

@@ -1,35 +0,0 @@
-/*#import "fmt.odin"
-
-thing :: proc() {
-	fmt.println("Hello1!")
-}*/
-
-
-#import "fmt.odin";
-
-main :: proc() {
-	fmt.println("hello, world!");
-}
-
-
-/*#import "fmt.odin" as .
-
-thing :: proc() {
-	println("Hello3!")
-}
-
-
-*/
-/*#import "fmt.odin" as _
-
-thing :: proc() {
-	// println("Hello4!")
-}
-*/
-
-/*
-#include "fmt.odin"
-
-thing :: proc() {
-	println("Hello5!")
-}*/

core/_preload.odin

@@ -1,665 +0,0 @@
-#shared_global_scope;
-
-#import "os.odin";
-#import "fmt.odin";
-#import "mem.odin";
-#import "utf8.odin";
-#import "hash.odin";
-
-// IMPORTANT NOTE(bill): `type_info` & `type_info_val` cannot be used within a
-// #shared_global_scope due to  the internals of the compiler.
-// This could change at a later date if the all these data structures are
-// implemented within the compiler rather than in this "preload" file
-
-
-// IMPORTANT NOTE(bill): Do not change the order of any of this data
-// The compiler relies upon this _exact_ order
-Type_Info_Member :: struct #ordered {
-	name:      string,     // can be empty if tuple
-	type_info: ^Type_Info,
-	offset:    int,        // offsets may not be used in tuples
-}
-Type_Info_Record :: struct #ordered {
-	fields:       []Type_Info_Member,
-	size:         int, // in bytes
-	align:        int, // in bytes
-	packed:       bool,
-	ordered:      bool,
-	custom_align: bool,
-}
-Type_Info_Enum_Value :: raw_union {
-	f: f64,
-	i: i64,
-}
-
-// NOTE(bill): This much the same as the compiler's
-Calling_Convention :: enum {
-	ODIN = 0,
-	C    = 1,
-	STD  = 2,
-	FAST = 3,
-}
-
-Type_Info :: union {
-	Named: struct #ordered {
-		name: string,
-		base: ^Type_Info, // This will _not_ be a Type_Info.Named
-	},
-	Integer: struct #ordered {
-		size:   int, // in bytes
-		signed: bool,
-	},
-	Float: struct #ordered {
-		size: int, // in bytes
-	},
-	String:  struct #ordered {},
-	Boolean: struct #ordered {},
-	Any:     struct #ordered {},
-	Pointer: struct #ordered {
-		elem: ^Type_Info, // nil -> rawptr
-	},
-	Procedure: struct #ordered {
-		params:     ^Type_Info, // Type_Info.Tuple
-		results:    ^Type_Info, // Type_Info.Tuple
-		variadic:   bool,
-		convention: Calling_Convention,
-	},
-	Array: struct #ordered {
-		elem:      ^Type_Info,
-		elem_size: int,
-		count:     int,
-	},
-	Dynamic_Array: struct #ordered {
-		elem:      ^Type_Info,
-		elem_size: int,
-	},
-	Slice: struct #ordered {
-		elem:      ^Type_Info,
-		elem_size: int,
-	},
-	Vector: struct #ordered {
-		elem:      ^Type_Info,
-		elem_size: int,
-		count:     int,
-		align:     int,
-	},
-	Tuple:     Type_Info_Record, // Only really used for procedures
-	Struct:    Type_Info_Record,
-	Union:     Type_Info_Record,
-	Raw_Union: Type_Info_Record,
-	Enum: struct #ordered {
-		base:   ^Type_Info,
-		names:  []string,
-		values: []Type_Info_Enum_Value,
-	},
-	Map: struct #ordered {
-		key:              ^Type_Info,
-		value:            ^Type_Info,
-		generated_struct: ^Type_Info,
-		count:            int, // == 0 if dynamic
-	},
-}
-
-// // NOTE(bill): only the ones that are needed (not all types)
-// // This will be set by the compiler
-// immutable __type_infos: []Type_Info;
-
-type_info_base :: proc(info: ^Type_Info) -> ^Type_Info {
-	if info == nil {
-		return nil;
-	}
-	base := info;
-	match i in base {
-	case Type_Info.Named:
-		base = i.base;
-	}
-	return base;
-}
-
-
-type_info_base_without_enum :: proc(info: ^Type_Info) -> ^Type_Info {
-	if info == nil {
-		return nil;
-	}
-	base := info;
-	match i in base {
-	case Type_Info.Named:
-		base = i.base;
-	case Type_Info.Enum:
-		base = i.base;
-	}
-	return base;
-}
-
-
-
-assume :: proc(cond: bool) #foreign __llvm_core "llvm.assume";
-
-__debug_trap       :: proc()        #foreign __llvm_core "llvm.debugtrap";
-__trap             :: proc()        #foreign __llvm_core "llvm.trap";
-read_cycle_counter :: proc() -> u64 #foreign __llvm_core "llvm.readcyclecounter";
-
-__cpuid :: proc(level: u32, sig: ^u32) -> i32 #foreign __llvm_core "__get_cpuid";
-
-
-
-
-
-// IMPORTANT NOTE(bill): Must be in this order (as the compiler relies upon it)
-Allocator_Mode :: enum u8 {
-	ALLOC,
-	FREE,
-	FREE_ALL,
-	RESIZE,
-}
-Allocator_Proc :: #type proc(allocator_data: rawptr, mode: Allocator_Mode,
-                             size, alignment: int,
-                             old_memory: rawptr, old_size: int, flags: u64) -> rawptr;
-Allocator :: struct #ordered {
-	procedure: Allocator_Proc,
-	data:      rawptr,
-}
-
-Context :: struct #ordered {
-	thread_id: int,
-
-	allocator: Allocator,
-
-	user_data:  rawptr,
-	user_index: int,
-}
-
-#thread_local __context: Context;
-
-
-DEFAULT_ALIGNMENT :: align_of([vector 4]f32);
-
-
-__check_context :: proc() {
-	c := ^__context;
-
-	if c.allocator.procedure == nil {
-		c.allocator = default_allocator();
-	}
-	if c.thread_id == 0 {
-		c.thread_id = os.current_thread_id();
-	}
-}
-
-alloc :: proc(size: int) -> rawptr #inline { return alloc_align(size, DEFAULT_ALIGNMENT); }
-
-alloc_align :: proc(size, alignment: int) -> rawptr #inline {
-	__check_context();
-	a := context.allocator;
-	return a.procedure(a.data, Allocator_Mode.ALLOC, size, alignment, nil, 0, 0);
-}
-
-free_ptr_with_allocator :: proc(a: Allocator, ptr: rawptr) #inline {
-	if ptr == nil {
-		return;
-	}
-	if a.procedure == nil {
-		return;
-	}
-	a.procedure(a.data, Allocator_Mode.FREE, 0, 0, ptr, 0, 0);
-}
-
-free_ptr :: proc(ptr: rawptr) #inline {
-	__check_context();
-	free_ptr_with_allocator(context.allocator, ptr);
-}
-
-free_all :: proc() #inline {
-	__check_context();
-	a := context.allocator;
-	a.procedure(a.data, Allocator_Mode.FREE_ALL, 0, 0, nil, 0, 0);
-}
-
-
-resize       :: proc(ptr: rawptr, old_size, new_size: int) -> rawptr #inline { return resize_align(ptr, old_size, new_size, DEFAULT_ALIGNMENT); }
-resize_align :: proc(ptr: rawptr, old_size, new_size, alignment: int) -> rawptr #inline {
-	__check_context();
-	a := context.allocator;
-	return a.procedure(a.data, Allocator_Mode.RESIZE, new_size, alignment, ptr, old_size, 0);
-}
-
-
-
-default_resize_align :: proc(old_memory: rawptr, old_size, new_size, alignment: int) -> rawptr {
-	if old_memory == nil {
-		return alloc_align(new_size, alignment);
-	}
-
-	if new_size == 0 {
-		free(old_memory);
-		return nil;
-	}
-
-	if new_size == old_size {
-		return old_memory;
-	}
-
-	new_memory := alloc_align(new_size, alignment);
-	if new_memory == nil {
-		return nil;
-	}
-
-	mem.copy(new_memory, old_memory, min(old_size, new_size));;
-	free(old_memory);
-	return new_memory;
-}
-
-
-default_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
-                               size, alignment: int,
-                               old_memory: rawptr, old_size: int, flags: u64) -> rawptr {
-	using Allocator_Mode;
-
-	match mode {
-	case ALLOC:
-		return os.heap_alloc(size);
-
-	case FREE:
-		os.heap_free(old_memory);
-		return nil;
-
-	case FREE_ALL:
-		// NOTE(bill): Does nothing
-
-	case RESIZE:
-		ptr := os.heap_resize(old_memory, size);
-		assert(ptr != nil);
-		return ptr;
-	}
-
-	return nil;
-}
-
-default_allocator :: proc() -> Allocator {
-	return Allocator{
-		procedure = default_allocator_proc,
-		data = nil,
-	};
-}
-
-
-
-
-
-
-
-
-
-
-
-__string_eq :: proc(a, b: string) -> bool {
-	if a.count != b.count {
-		return false;
-	}
-	if a.data == b.data {
-		return true;
-	}
-	return __string_cmp(a, b) == 0;
-}
-
-__string_cmp :: proc(a, b: string) -> int {
-	return mem.compare(cast([]byte)a, cast([]byte)b);
-}
-
-__string_ne :: proc(a, b: string) -> bool #inline { return !__string_eq(a, b); }
-__string_lt :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) < 0; }
-__string_gt :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) > 0; }
-__string_le :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) <= 0; }
-__string_ge :: proc(a, b: string) -> bool #inline { return __string_cmp(a, b) >= 0; }
-
-
-__assert :: proc(file: string, line, column: int, msg: string) #inline {
-	fmt.fprintf(os.stderr, "%s(%d:%d) Runtime assertion: %s\n",
-	            file, line, column, msg);
-	__debug_trap();
-}
-
-__bounds_check_error :: proc(file: string, line, column: int, index, count: int) {
-	if 0 <= index && index < count {
-		return;
-	}
-	fmt.fprintf(os.stderr, "%s(%d:%d) Index %d is out of bounds range 0..<%d\n",
-	            file, line, column, index, count);
-	__debug_trap();
-}
-
-__slice_expr_error :: proc(file: string, line, column: int, low, high: int) {
-	if 0 <= low && low <= high {
-		return;
-	}
-	fmt.fprintf(os.stderr, "%s(%d:%d) Invalid slice indices: [%d:%d]\n",
-	            file, line, column, low, high);
-	__debug_trap();
-}
-__substring_expr_error :: proc(file: string, line, column: int, low, high: int) {
-	if 0 <= low && low <= high {
-		return;
-	}
-	fmt.fprintf(os.stderr, "%s(%d:%d) Invalid substring indices: [%d:%d]\n",
-	            file, line, column, low, high);
-	__debug_trap();
-}
-
-__string_decode_rune :: proc(s: string) -> (rune, int) #inline {
-	return utf8.decode_rune(s);
-}
-
-
-Raw_Any :: struct #ordered {
-	type_info: ^Type_Info,
-	data:      rawptr,
-}
-
-Raw_String :: struct #ordered {
-	data:  ^byte,
-	count: int,
-};
-
-Raw_Slice :: struct #ordered {
-	data:  rawptr,
-	count: int,
-};
-
-Raw_Dynamic_Array :: struct #ordered {
-	data:      rawptr,
-	count:     int,
-	capacity:  int,
-	allocator: Allocator,
-};
-
-Raw_Dynamic_Map :: struct #ordered {
-	hashes:  [...]int,
-	entries: Raw_Dynamic_Array,
-};
-
-
-
-__dynamic_array_reserve :: proc(array_: rawptr, elem_size, elem_align: int, capacity: int) -> bool {
-	array := cast(^Raw_Dynamic_Array)array_;
-
-	if capacity <= array.capacity {
-		return true;
-	}
-
-	__check_context();
-	if array.allocator.procedure == nil {
-		array.allocator = context.allocator;
-	}
-	assert(array.allocator.procedure != nil);
-
-	old_size  := array.capacity * elem_size;
-	new_size  := capacity * elem_size;
-	allocator := array.allocator;
-
-	new_data := allocator.procedure(allocator.data, Allocator_Mode.RESIZE, new_size, elem_align, array.data, old_size, 0);
-	if new_data == nil {
-		return false;
-	}
-
-	array.data = new_data;
-	array.capacity = capacity;
-	return true;
-}
-
-
-__dynamic_array_append :: proc(array_: rawptr, elem_size, elem_align: int,
-                               items: rawptr, item_count: int) -> int {
-	array := cast(^Raw_Dynamic_Array)array_;
-
-	if item_count <= 0 || items == nil {
-		return array.count;
-	}
-
-
-	ok := true;
-	if array.capacity <= array.count+item_count {
-		capacity := 2 * array.capacity + max(8, item_count);
-		ok = __dynamic_array_reserve(array, elem_size, elem_align, capacity);
-	}
-	if !ok {
-		// TODO(bill): Better error handling for failed reservation
-		return array.count;
-	}
-	data := cast(^byte)array.data;
-	assert(data != nil);
-	mem.copy(data + (elem_size*array.count), items, elem_size * item_count);
-	array.count += item_count;
-	return array.count;
-}
-
-__dynamic_array_append_nothing :: proc(array_: rawptr, elem_size, elem_align: int) -> int {
-	array := cast(^Raw_Dynamic_Array)array_;
-
-	ok := true;
-	if array.capacity <= array.count+1 {
-		capacity := 2 * array.capacity + max(8, 1);
-		ok = __dynamic_array_reserve(array, elem_size, elem_align, capacity);
-	}
-	if !ok {
-		// TODO(bill): Better error handling for failed reservation
-		return array.count;
-	}
-	data := cast(^byte)array.data;
-	assert(data != nil);
-	mem.zero(data + (elem_size*array.count), elem_size);
-	array.count += 1;
-	return array.count;
-}
-
-
-__default_hash :: proc(data: []byte) -> u64 {
-	return hash.fnv64a(data);
-}
-__default_hash_string :: proc(s: string) -> u64 {
-	return __default_hash(cast([]byte)s);
-}
-
-__Map_Key :: struct #ordered {
-	hash: u64,
-	str:  string,
-}
-
-__Map_Find_Result :: struct #ordered {
-	hash_index:  int,
-	entry_prev:  int,
-	entry_index: int,
-}
-
-__Map_Entry_Header :: struct #ordered {
-	key:  __Map_Key,
-	next: int,
-/*
-	value: Value_Type,
-*/
-}
-
-__Map_Header :: struct #ordered {
-	m:             ^Raw_Dynamic_Map,
-	is_key_string: bool,
-	entry_size:    int,
-	entry_align:   int,
-	value_offset:  int,
-}
-
-__dynamic_map_reserve :: proc(using header: __Map_Header, capacity: int) -> bool {
-	h := __dynamic_array_reserve(^m.hashes, size_of(int), align_of(int), capacity);
-	e := __dynamic_array_reserve(^m.entries, entry_size, entry_align,    capacity);
-	return h && e;
-}
-
-__dynamic_map_rehash :: proc(using header: __Map_Header, new_count: int) {
-	new_header := header;
-	nm: Raw_Dynamic_Map;
-	new_header.m = ^nm;
-
-	reserve(nm.hashes, new_count);
-	nm.hashes.count = nm.hashes.capacity;
-	__dynamic_array_reserve(^nm.entries, entry_size, entry_align, m.entries.count);
-	for _, i in nm.hashes {
-		nm.hashes[i] = -1;
-	}
-
-	for i := 0; i < nm.entries.count; i += 1 {
-		entry_header := __dynamic_map_get_entry(new_header, i);
-		data := cast(^byte)entry_header;
-
-		if nm.hashes.count == 0 {
-			__dynamic_map_grow(new_header);
-		}
-
-		fr := __dynamic_map_find(new_header, entry_header.key);
-		j := __dynamic_map_add_entry(new_header, entry_header.key);
-		if fr.entry_prev < 0 {
-			nm.hashes[fr.hash_index] = j;
-		} else {
-			e := __dynamic_map_get_entry(new_header, fr.entry_prev);
-			e.next = j;
-		}
-
-		e := __dynamic_map_get_entry(new_header, j);
-		e.next = fr.entry_index;
-		ndata := cast(^byte)e;
-		mem.copy(ndata+value_offset, data+value_offset, entry_size-value_offset);
-		if __dynamic_map_full(new_header) {
-			__dynamic_map_grow(new_header);
-		}
-	}
-	free_ptr_with_allocator(header.m.hashes.allocator,  header.m.hashes.data);
-	free_ptr_with_allocator(header.m.entries.allocator, header.m.entries.data);
-	header.m^ = nm;
-}
-
-__dynamic_map_get :: proc(h: __Map_Header, key: __Map_Key) -> rawptr {
-	index := __dynamic_map_find(h, key).entry_index;
-	if index >= 0 {
-		data := cast(^byte)__dynamic_map_get_entry(h, index);
-		val := data + h.value_offset;
-		return val;
-	}
-	return nil;
-}
-
-__dynamic_map_set :: proc(using h: __Map_Header, key: __Map_Key, value: rawptr) {
-	index: int;
-
-	if m.hashes.count == 0 {
-		__dynamic_map_grow(h);
-	}
-	fr := __dynamic_map_find(h, key);
-	if fr.entry_index >= 0 {
-		index = fr.entry_index;
-	} else {
-		index = __dynamic_map_add_entry(h, key);
-		if fr.entry_prev >= 0 {
-			entry := __dynamic_map_get_entry(h, fr.entry_prev);
-			entry.next = index;
-		} else {
-			m.hashes[fr.hash_index] = index;
-		}
-	}
-	{
-		data := cast(^byte)__dynamic_map_get_entry(h, index);
-		val := data+value_offset;
-		mem.copy(val, value, entry_size-value_offset);
-	}
-
-	if __dynamic_map_full(h) {
-		__dynamic_map_grow(h);
-	}
-}
-
-
-__dynamic_map_grow :: proc(using h: __Map_Header) {
-	new_count := 2*m.entries.count + 8;
-	__dynamic_map_rehash(h, new_count);
-}
-
-__dynamic_map_full :: proc(using h: __Map_Header) -> bool {
-	return cast(int)(0.75 * cast(f64)m.hashes.count) <= m.entries.count;
-}
-
-
-__dynamic_map_hash_equal :: proc(h: __Map_Header, a, b: __Map_Key) -> bool {
-	if a.hash == b.hash {
-		if h.is_key_string {
-			return a.str == b.str;
-		}
-		return true;
-	}
-	return false;
-}
-
-__dynamic_map_find :: proc(using h: __Map_Header, key: __Map_Key) -> __Map_Find_Result {
-	fr := __Map_Find_Result{-1, -1, -1};
-	if m.hashes.count > 0 {
-		fr.hash_index = cast(int)(key.hash % cast(u64)m.hashes.count);
-		fr.entry_index = m.hashes[fr.hash_index];
-		for fr.entry_index >= 0 {
-			entry := __dynamic_map_get_entry(h, fr.entry_index);
-			if __dynamic_map_hash_equal(h, entry.key, key) {
-				return fr;
-			}
-			fr.entry_prev = fr.entry_index;
-			fr.entry_index = entry.next;
-		}
-	}
-	return fr;
-}
-
-__dynamic_map_add_entry :: proc(using h: __Map_Header, key: __Map_Key) -> int {
-	prev := m.entries.count;
-	c := __dynamic_array_append_nothing(^m.entries, entry_size, entry_align);
-	if c != prev {
-		end := __dynamic_map_get_entry(h, c-1);
-		end.key = key;
-		end.next = -1;
-	}
-	return prev;
-}
-
-
-__dynamic_map_remove :: proc(using h: __Map_Header, key: __Map_Key) {
-	fr := __dynamic_map_find(h, key);
-	if fr.entry_index >= 0 {
-		__dynamic_map_erase(h, fr);
-	}
-}
-
-__dynamic_map_get_entry :: proc(using h: __Map_Header, index: int) -> ^__Map_Entry_Header {
-	data := cast(^byte)m.entries.data + index*entry_size;
-	return cast(^__Map_Entry_Header)data;
-}
-
-__dynamic_map_erase :: proc(using h: __Map_Header, fr: __Map_Find_Result) {
-	if fr.entry_prev < 0 {
-		m.hashes[fr.hash_index] = __dynamic_map_get_entry(h, fr.entry_index).next;
-	} else {
-		__dynamic_map_get_entry(h, fr.entry_prev).next = __dynamic_map_get_entry(h, fr.entry_index).next;
-	}
-
-	if fr.entry_index == m.entries.count-1 {
-		m.entries.count -= 1;
-	}
-	mem.copy(__dynamic_map_get_entry(h, fr.entry_index), __dynamic_map_get_entry(h, m.entries.count-1), entry_size);
-	last := __dynamic_map_find(h, __dynamic_map_get_entry(h, fr.entry_index).key);
-	if last.entry_prev >= 0 {
-		__dynamic_map_get_entry(h, last.entry_prev).next = fr.entry_index;
-	} else {
-		m.hashes[last.hash_index] = fr.entry_index;
-	}
-}
-
-
-__print_ti_ptr :: proc(ti: ^Type_Info) {
-	fmt.println(ti);
-	match e in ti {
-	case Type_Info.Enum:
-		fmt.println(e.names);
-	}
-}
-

core/_soft_numbers.odin

@@ -1,158 +0,0 @@
-#shared_global_scope;
-
-
-// import "fmt.odin";
-
-// proc __u128_mod(a, b: u128) -> u128 #link_name "__umodti3" {
-// 	var _, r := __u128_quo_mod(a, b)
-// 	return r
-// }
-
-// proc __u128_quo(a, b: u128) -> u128 #link_name "__udivti3" {
-// 	var n, _ := __u128_quo_mod(a, b)
-// 	return n
-// }
-
-// proc __i128_mod(a, b: i128) -> i128 #link_name "__modti3" {
-// 	var _, r := __i128_quo_mod(a, b)
-// 	return r
-// }
-
-// proc __i128_quo(a, b: i128) -> i128 #link_name "__divti3" {
-// 	var n, _ := __i128_quo_mod(a, b)
-// 	return n
-// }
-
-// proc __i128_quo_mod(a, b: i128) -> (i128, i128) #link_name "__divmodti4" {
-// 	var s := b >> 127
-// 	b = (b ~ s) - s
-// 	s = a >> 127
-// 	a = (a ~ s) - s
-
-// 	var n, r := __u128_quo_mod(a as u128, b as u128)
-// 	return (n as i128 ~ s) - s, (r as i128 ~ s) - s
-// }
-
-
-// proc __u128_quo_mod(a, b: u128) -> (u128, u128) #link_name "__udivmodti4" {
-// 	proc clz(x: u64) -> u64 {
-// 		proc clz_u64(x: u64, is_zero_undef: bool) -> u64 #foreign "llvm.ctlz.i64"
-// 		return clz_u64(x, false)
-// 	}
-// 	proc ctz(x: u64) -> u64 {
-// 		proc ctz_u64(x: u64, is_zero_undef: bool) -> u64 #foreign "llvm.cttz.i64"
-// 		return ctz_u64(x, false)
-// 	}
-
-
-// 	u128_lo_hi :: raw_union {
-// 		all: u128
-// 		using _lohi: struct {lo, hi: u64;}
-// 	}
-
-// 	n, d, q, r: u128_lo_hi
-// 	sr: u64
-
-// 	n.all = a
-// 	d.all = b
-
-// 	if n.hi == 0 {
-// 		if d.hi == 0 {
-// 			return (n.lo / d.lo) as u128, (n.lo % d.lo) as u128
-// 		}
-// 		return 0, n.lo as u128
-// 	}
-// 	if d.lo == 0 {
-// 		if d.hi == 0 {
-// 			return (n.hi / d.lo) as u128, (n.hi % d.lo) as u128
-// 		}
-// 		if n.lo == 0 {
-// 			r.hi = n.hi % d.hi
-// 			r.lo = 0
-// 			return (n.hi / d.hi) as u128, r.all
-// 		}
-// 		if (d.hi & (d.hi-1)) == 0 {
-// 			r.lo = n.lo
-// 			r.hi = n.hi & (d.hi-1)
-// 			return (n.hi >> ctz(d.hi)) as u128, r.all
-// 		}
-
-// 		sr = clz(d.hi) - clz(n.hi)
-// 		if sr > 64 - 2 {
-// 			return 0, n.all
-// 		}
-// 		sr++
-// 		q.lo = 0
-// 		q.hi = n.lo << (64-sr)
-// 		r.hi = n.hi >> sr
-// 		r.lo = (n.hi << (64-sr)) | (n.lo >> sr)
-// 	} else {
-// 		if d.hi == 0 {
-// 			if (d.lo & (d.lo - 1)) == 0 {
-// 				var rem := (n.lo % (d.lo - 1)) as u128
-// 				if d.lo == 1 {
-// 					return n.all, rem
-// 				}
-// 				sr = ctz(d.lo)
-// 				q.hi = n.hi >> sr
-// 				q.lo = (n.hi << (64-sr)) | (n.lo >> sr)
-// 				return q.all, rem
-// 			}
-
-// 			sr = 1 + 64 + clz(d.lo) - clz(n.hi)
-
-// 			q.all = n.all << (128-sr)
-// 			r.all = n.all >> sr
-// 			if sr == 64 {
-// 				q.lo = 0
-// 				q.hi = n.lo
-// 				r.hi = 0
-// 				r.lo = n.hi
-// 			} else if sr < 64 {
-// 				q.lo = 0
-// 				q.hi = n.lo << (64-sr)
-// 				r.hi = n.hi >> sr
-// 				r.lo = (n.hi << (64-sr)) | (n.lo >> sr)
-// 			} else {
-// 				q.lo = n.lo << (128-sr)
-// 				q.hi = (n.hi << (128-sr)) | (n.lo >> (sr-64))
-// 				r.hi = 0
-// 				r.lo = n.hi >> (sr-64)
-// 			}
-// 		} else {
-// 			sr = clz(d.hi) - clz(n.hi)
-// 			if sr > 64-1 {
-// 				return 0, n.all
-// 			}
-// 			sr++
-// 			q.lo = 0
-// 			q.hi = n.lo << (64-sr)
-// 			r.all = n.all >> sr
-// 			if sr < 64 {
-// 				r.hi = n.hi >> sr
-// 				r.lo = (n.hi << (64-sr)) | (n.lo >> sr)
-// 			} else {
-// 				r.hi = 0
-// 				r.lo = n.hi
-// 			}
-// 		}
-// 	}
-
-// 	carry: u64
-// 	for ; sr > 0; sr-- {
-// 		r.hi = (r.hi << 1) | (r.lo >> (64-1))
-// 		r.lo = (r.lo << 1) | (r.hi >> (64-1))
-// 		q.hi = (q.hi << 1) | (q.lo >> (64-1))
-// 		q.lo = (q.lo << 1) | carry
-
-// 		carry = 0
-// 		if r.all >= d.all {
-// 			r.all -= d.all
-// 			carry = 1
-// 		}
-// 	}
-
-// 	q.all = (q.all << 1) | (carry as u128)
-// 	return q.all, r.all
-// }
-

core/atomic.odin

@@ -1,100 +0,0 @@
-// TODO(bill): Use assembly instead here to implement atomics
-// Inline vs external file?
-
-#import win32 "sys/windows.odin" when ODIN_OS == "windows";
-_ := compile_assert(ODIN_ARCH == "amd64"); // TODO(bill): x86 version
-
-
-yield_thread :: proc() { win32.mm_pause(); }
-mfence       :: proc() { win32.ReadWriteBarrier(); }
-sfence       :: proc() { win32.WriteBarrier(); }
-lfence       :: proc() { win32.ReadBarrier(); }
-
-
-load :: proc(a: ^i32) -> i32 {
-	return a^;
-}
-store :: proc(a: ^i32, value: i32) {
-	a^ = value;
-}
-compare_exchange :: proc(a: ^i32, expected, desired: i32) -> i32 {
-	return win32.InterlockedCompareExchange(a, desired, expected);
-}
-exchanged :: proc(a: ^i32, desired: i32) -> i32 {
-	return win32.InterlockedExchange(a, desired);
-}
-fetch_add :: proc(a: ^i32, operand: i32) -> i32 {
-	return win32.InterlockedExchangeAdd(a, operand);
-
-}
-fetch_and :: proc(a: ^i32, operand: i32) -> i32 {
-	return win32.InterlockedAnd(a, operand);
-}
-fetch_or :: proc(a: ^i32, operand: i32) -> i32 {
-	return win32.InterlockedOr(a, operand);
-}
-spin_lock :: proc(a: ^i32, time_out: int) -> bool { // NOTE(bill) time_out = -1 as default
-	old_value := compare_exchange(a, 1, 0);
-	counter := 0;
-	for old_value != 0 && (time_out < 0 || counter < time_out) {
-		counter += 1;
-		yield_thread();
-		old_value = compare_exchange(a, 1, 0);
-		mfence();
-	}
-	return old_value == 0;
-}
-spin_unlock :: proc(a: ^i32) {
-	store(a, 0);
-	mfence();
-}
-try_acquire_lock :: proc(a: ^i32) -> bool {
-	yield_thread();
-	old_value := compare_exchange(a, 1, 0);
-	mfence();
-	return old_value == 0;
-}
-
-
-load :: proc(a: ^i64) -> i64 {
-	return a^;
-}
-store :: proc(a: ^i64, value: i64) {
-	a^ = value;
-}
-compare_exchange :: proc(a: ^i64, expected, desired: i64) -> i64 {
-	return win32.InterlockedCompareExchange64(a, desired, expected);
-}
-exchanged :: proc(a: ^i64, desired: i64) -> i64 {
-	return win32.InterlockedExchange64(a, desired);
-}
-fetch_add :: proc(a: ^i64, operand: i64) -> i64 {
-	return win32.InterlockedExchangeAdd64(a, operand);
-}
-fetch_and :: proc(a: ^i64, operand: i64) -> i64 {
-	return win32.InterlockedAnd64(a, operand);
-}
-fetch_or :: proc(a: ^i64, operand: i64) -> i64 {
-	return win32.InterlockedOr64(a, operand);
-}
-spin_lock :: proc(a: ^i64, time_out: int) -> bool { // NOTE(bill) time_out = -1 as default
-	old_value := compare_exchange(a, 1, 0);
-	counter := 0;
-	for old_value != 0 && (time_out < 0 || counter < time_out) {
-		counter += 1;
-		yield_thread();
-		old_value = compare_exchange(a, 1, 0);
-		mfence();
-	}
-	return old_value == 0;
-}
-spin_unlock :: proc(a: ^i64) {
-	store(a, 0);
-	mfence();
-}
-try_acquire_lock :: proc(a: ^i64) -> bool {
-	yield_thread();
-	old_value := compare_exchange(a, 1, 0);
-	mfence();
-	return old_value == 0;
-}

core/c/c.odin

@@ -0,0 +1,33 @@
+package c
+
+import b "core:builtin"
+
+CHAR_BIT :: 8;
+
+bool   :: b.bool;
+char   :: b.u8;
+byte   :: b.byte;
+schar  :: b.i8;
+uchar  :: b.u8;
+short  :: b.i16;
+ushort :: b.u16;
+int    :: b.i32;
+uint   :: b.u32;
+
+long  :: (ODIN_OS == "windows" || size_of(b.rawptr) == 4) ? b.i32 : b.i64;
+ulong :: (ODIN_OS == "windows" || size_of(b.rawptr) == 4) ? b.u32 : b.u64;
+
+longlong       :: b.i64;
+ulonglong      :: b.u64;
+float          :: b.f32;
+double         :: b.f64;
+complex_float  :: b.complex64;
+complex_double :: b.complex128;
+
+#assert(size_of(b.uintptr) == size_of(b.int));
+
+size_t    :: b.uint;
+ssize_t   :: b.int;
+ptrdiff_t :: b.int;
+uintptr_t :: b.uintptr;
+intptr_t  :: b.int;

core/decimal/decimal.odin

@@ -0,0 +1,264 @@
+// Multiple precision decimal numbers
+// NOTE: This is only for floating point printing and nothing else
+package decimal
+
+Decimal :: struct {
+	digits:        [384]byte, // big-endian digits
+	count:         int,
+	decimal_point: int,
+	neg, trunc:    bool,
+}
+
+decimal_to_string :: proc(buf: []byte, a: ^Decimal) -> string {
+	digit_zero :: proc(buf: []byte) -> int {
+		for _, i in buf do buf[i] = '0';
+		return len(buf);
+	}
+
+
+	n := 10 + a.count + abs(a.decimal_point);
+
+	// TODO(bill): make this work with a buffer that's not big enough
+	assert(len(buf) >= n);
+	buf = buf[0:n];
+
+	if a.count == 0 {
+		buf[0] = '0';
+		return string(buf[0:1]);
+	}
+
+	w := 0;
+	if a.decimal_point <= 0 {
+		buf[w] = '0'; w += 1;
+		buf[w] = '.'; w += 1;
+		w += digit_zero(buf[w : w-a.decimal_point]);
+		w += copy(buf[w:], a.digits[0:a.count]);
+	} else if a.decimal_point < a.count {
+		w += copy(buf[w:], a.digits[0:a.decimal_point]);
+		buf[w] = '.'; w += 1;
+		w += copy(buf[w:], a.digits[a.decimal_point : a.count]);
+	} else {
+		w += copy(buf[w:], a.digits[0:a.count]);
+		w += digit_zero(buf[w : w+a.decimal_point-a.count]);
+	}
+
+	return string(buf[0:w]);
+}
+
+// trim trailing zeros
+trim :: proc(a: ^Decimal) {
+	for a.count > 0 && a.digits[a.count-1] == '0' {
+		a.count -= 1;
+	}
+	if a.count == 0 {
+		a.decimal_point = 0;
+	}
+}
+
+
+assign :: proc(a: ^Decimal, i: u64) {
+	buf: [64]byte;
+	n := 0;
+	for i > 0 {
+		j := i/10;
+		i -= 10*j;
+		buf[n] = byte('0'+i);
+		n += 1;
+		i = j;
+	}
+
+	a.count = 0;
+	for n -= 1; n >= 0; n -= 1 {
+		a.digits[a.count] = buf[n];
+		a.count += 1;
+	}
+	a.decimal_point = a.count;
+	trim(a);
+}
+
+
+
+shift_right :: proc(a: ^Decimal, k: uint) {
+	r := 0; // read index
+	w := 0; // write index
+
+	n: uint;
+	for ; n>>k == 0; r += 1 {
+		if r >= a.count {
+			if n == 0 {
+				// Just in case
+				a.count = 0;
+				return;
+			}
+			for n>>k == 0 {
+				n = n * 10;
+				r += 1;
+			}
+			break;
+		}
+		c := uint(a.digits[r]);
+		n = n*10 + c - '0';
+	}
+	a.decimal_point -= r-1;
+
+	mask: uint = (1<<k) - 1;
+
+	for ; r < a.count; r += 1 {
+		c := uint(a.digits[r]);
+		dig := n>>k;
+		n &= mask;
+		a.digits[w] = byte('0' + dig);
+		w += 1;
+		n = n*10 + c - '0';
+	}
+
+	for n > 0 {
+		dig := n>>k;
+		n &= mask;
+		if w < len(a.digits) {
+			a.digits[w] = byte('0' + dig);
+			w += 1;
+		} else if dig > 0 {
+			a.trunc = true;
+		}
+		n *= 10;
+	}
+
+
+	a.count = w;
+	trim(a);
+}
+
+shift_left :: proc(a: ^Decimal, k: uint) {
+	// NOTE(bill): used to determine buffer size required for the decimal from the binary shift
+	// 'k' means `1<<k` == `2^k` which equates to roundup(k*log10(2)) digits required
+	log10_2 :: 0.301029995663981195213738894724493026768189881462108541310;
+	capacity := int(f64(k)*log10_2 + 1);
+
+	r := a.count;          // read index
+	w := a.count+capacity; // write index
+
+	d := len(a.digits);
+
+	n: uint;
+	for r -= 1; r >= 0; r -= 1 {
+		n += (uint(a.digits[r]) - '0') << k;
+		quo := n/10;
+		rem := n - 10*quo;
+		w -= 1;
+		if w < d {
+			a.digits[w] = byte('0' + rem);
+		} else if rem != 0 {
+			a.trunc = true;
+		}
+		n = quo;
+	}
+
+	for n > 0 {
+		quo := n/10;
+		rem := n - 10*quo;
+		w -= 1;
+		if w < d {
+			a.digits[w] = byte('0' + rem);
+		} else if rem != 0 {
+			a.trunc = true;
+		}
+		n = quo;
+	}
+
+	// NOTE(bill): Remove unused buffer size
+	assert(w >= 0);
+	capacity -= w;
+
+	a.count = min(a.count+capacity, d);
+	a.decimal_point += capacity;
+	trim(a);
+}
+
+shift :: proc(a: ^Decimal, k: int) {
+	uint_size :: 8*size_of(uint);
+	max_shift :: uint_size-4;
+
+	switch {
+	case a.count == 0:
+		// no need to update
+	case k > 0:
+		for k > max_shift {
+			shift_left(a, max_shift);
+			k -= max_shift;
+		}
+		shift_left(a, uint(k));
+
+
+	case k < 0:
+		for k < -max_shift {
+			shift_right(a, max_shift);
+			k += max_shift;
+		}
+		shift_right(a, uint(-k));
+	}
+}
+
+can_round_up :: proc(a: ^Decimal, nd: int) -> bool {
+	if nd < 0 || nd >= a.count { return false ; }
+	if a.digits[nd] == '5' && nd+1 == a.count {
+		if a.trunc do return true;
+		return nd > 0 && (a.digits[nd-1]-'0')%2 != 0;
+	}
+
+	return a.digits[nd] >= '5';
+}
+
+round :: proc(a: ^Decimal, nd: int) {
+	if nd < 0 || nd >= a.count { return; }
+	if can_round_up(a, nd) {
+		round_up(a, nd);
+	} else {
+		round_down(a, nd);
+	}
+}
+
+round_up :: proc(a: ^Decimal, nd: int) {
+	if nd < 0 || nd >= a.count { return; }
+
+	for i := nd-1; i >= 0; i -= 1 {
+		if c := a.digits[i]; c < '9' {
+			a.digits[i] += 1;
+			a.count = i+1;
+			return;
+		}
+	}
+
+	// Number is just 9s
+	a.digits[0] = '1';
+	a.count = 1;
+	a.decimal_point += 1;
+}
+
+round_down :: proc(a: ^Decimal, nd: int) {
+	if nd < 0 || nd >= a.count { return; }
+	a.count = nd;
+	trim(a);
+}
+
+
+// Extract integer part, rounded appropriately. There are no guarantees about overflow.
+rounded_integer :: proc(a: ^Decimal) -> u64 {
+	if a.decimal_point > 20 {
+		return 0xffff_ffff_ffff_ffff;
+	}
+	i: int = 0;
+	n: u64 = 0;
+	m := min(a.decimal_point, a.count);
+	for ; i < m; i += 1 {
+		n = n*10 + u64(a.digits[i]-'0');
+	}
+	for ; i < a.decimal_point; i += 1 {
+		n *= 10;
+	}
+	if can_round_up(a, a.decimal_point) {
+		n += 1;
+	}
+	return n;
+}
+

core/dynlib/lib.odin

@@ -0,0 +1,3 @@
+package dynlib
+
+Library :: opaque rawptr;

core/dynlib/lib_windows.odin

@@ -0,0 +1,24 @@
+package dynlib
+
+import "core:sys/win32"
+import "core:strings"
+
+load_library :: proc(path: string, global_symbols := false) -> (Library, bool) {
+	// NOTE(bill): 'global_symbols' is here only for consistency with POSIX which has RTLD_GLOBAL
+
+	wide_path := win32.utf8_to_wstring(path, context.temp_allocator);
+	handle := cast(Library)win32.load_library_w(wide_path);
+	return handle, handle != nil;
+}
+
+unload_library :: proc(library: Library) -> bool {
+	ok := win32.free_library(cast(win32.Hmodule)library);
+	return bool(ok);
+}
+
+symbol_address :: proc(library: Library, symbol: string) -> (ptr: rawptr, found: bool) {
+	c_str := strings.clone_to_cstring(symbol, context.temp_allocator);
+	ptr = win32.get_proc_address(cast(win32.Hmodule)library, c_str);
+	found == ptr != nil;
+	return;
+}

core/encoding/cel/cel.odin

@@ -0,0 +1,838 @@
+package cel;
+
+import "core:fmt"
+import "core:strconv"
+import "core:unicode/utf8"
+import "core:strings"
+
+Array :: []Value;
+Dict  :: map[string]Value;
+Nil_Value :: struct{};
+
+Value :: union {
+	Nil_Value,
+	bool, i64, f64, string,
+	Array, Dict,
+}
+
+Parser :: struct {
+	tokens:           [dynamic]Token,
+	prev_token:       Token,
+	curr_token:       Token,
+	curr_token_index: int,
+
+	allocated_strings: [dynamic]string,
+
+	error_count: int,
+
+	root: Dict,
+	dict_stack: [dynamic]^Dict, // NOTE: Pointers may be stored on the stack
+}
+
+
+print_value :: proc(value: Value, pretty := true, indent := 0) {
+	print_indent :: proc(indent: int) {
+		for _ in 0..<indent do fmt.print("\t");
+	}
+
+	switch v in value {
+	case bool: fmt.print(v);
+	case i64: fmt.print(v);
+	case f64: fmt.print(v);
+	case string: fmt.print(v);
+	case Array:
+		fmt.print("[");
+		if pretty do fmt.println();
+		for e, i in v {
+			if pretty {
+				print_indent(indent+1);
+				print_value(e, pretty, indent+1);
+				fmt.println(",");
+			} else {
+				if i > 0 do fmt.print(", ");
+				print_value(e);
+			}
+		}
+		if pretty do print_indent(indent);
+		fmt.print("]");
+	case Dict:
+		fmt.print("{");
+		if pretty do fmt.println();
+
+		i := 0;
+		for name, val in v {
+			if pretty {
+				print_indent(indent+1);
+				fmt.printf("%s = ", name);
+				print_value(val, pretty, indent+1);
+				fmt.println(",");
+			} else {
+				if i > 0 do fmt.print(", ");
+				fmt.printf("%s = ", name);
+				print_value(val, pretty, indent+1);
+				i += 1;
+			}
+		}
+
+		if pretty do print_indent(indent);
+		fmt.print("}");
+	case:
+		fmt.print("nil");
+	case Nil_Value:
+		fmt.print("nil");
+	}
+}
+print :: proc(p: ^Parser, pretty := false) {
+	for name, val in p.root {
+		fmt.printf("%s = ", name);
+		print_value(val, pretty);
+		fmt.println(";");
+	}
+}
+
+create_from_string :: proc(src: string) -> (^Parser, bool) {
+	return init(cast([]byte)src);
+}
+
+
+init :: proc(src: []byte) -> (^Parser, bool) {
+	t: Tokenizer;
+	tokenizer_init(&t, src);
+	return create_from_tokenizer(&t);
+}
+
+
+create_from_tokenizer :: proc(t: ^Tokenizer) -> (^Parser, bool) {
+	p := new(Parser);
+	for {
+		tok := scan(t);
+		if tok.kind == Kind.Illegal {
+			return p, false;
+		}
+		append(&p.tokens, tok);
+		if tok.kind == Kind.EOF {
+			break;
+		}
+	}
+
+	if t.error_count > 0 {
+		return p, false;
+	}
+
+	if len(p.tokens) == 0 {
+		tok := Token{kind = Kind.EOF};
+		tok.line, tok.column = 1, 1;
+		append(&p.tokens, tok);
+		return p, true;
+	}
+
+	p.curr_token_index = 0;
+	p.prev_token = p.tokens[p.curr_token_index];
+	p.curr_token = p.tokens[p.curr_token_index];
+
+	p.root = Dict{};
+	p.dict_stack = make([dynamic]^Dict, 0, 4);
+	append(&p.dict_stack, &p.root);
+
+	for p.curr_token.kind != Kind.EOF &&
+	    p.curr_token.kind != Kind.Illegal &&
+	    p.curr_token_index < len(p.tokens) {
+		if !parse_assignment(p) {
+			break;
+		}
+	}
+
+	return p, true;
+}
+
+destroy :: proc(p: ^Parser) {
+	destroy_value :: proc(value: Value) {
+		switch v in value {
+		case Array:
+			for elem in v do destroy_value(elem);
+			delete(v);
+
+		case Dict:
+			for _, dv in v do destroy_value(dv);
+			delete(v);
+		}
+	}
+
+	delete(p.tokens);
+	for s in p.allocated_strings do delete(s);
+	delete(p.allocated_strings);
+	delete(p.dict_stack);
+
+	destroy_value(p.root);
+	free(p);
+}
+
+error :: proc(p: ^Parser, pos: Pos, msg: string, args: ..any) {
+	fmt.printf_err("%s(%d:%d) Error: ", pos.file, pos.line, pos.column);
+	fmt.printf_err(msg, ..args);
+	fmt.println_err();
+
+	p.error_count += 1;
+}
+
+next_token :: proc(p: ^Parser) -> Token {
+	p.prev_token = p.curr_token;
+	prev := p.prev_token;
+
+	if p.curr_token_index+1 < len(p.tokens) {
+		p.curr_token_index += 1;
+		p.curr_token = p.tokens[p.curr_token_index];
+		return prev;
+	}
+	p.curr_token_index = len(p.tokens);
+	p.curr_token = p.tokens[p.curr_token_index-1];
+	error(p, prev.pos, "Token is EOF");
+	return prev;
+}
+
+unquote_char :: proc(s: string, quote: byte) -> (r: rune, multiple_bytes: bool, tail_string: string, success: bool) {
+	hex_to_int :: proc(c: byte) -> int {
+		switch c {
+		case '0'..'9': return int(c-'0');
+		case 'a'..'f': return int(c-'a')+10;
+		case 'A'..'F': return int(c-'A')+10;
+		}
+		return -1;
+	}
+	w: int;
+
+	if s[0] == quote && quote == '"' {
+		return;
+	} else if s[0] >= 0x80 {
+		r, w = utf8.decode_rune_in_string(s);
+		return r, true, s[w:], true;
+	} else if s[0] != '\\' {
+		return rune(s[0]), false, s[1:], true;
+	}
+
+	if len(s) <= 1 {
+		return;
+	}
+	c := s[1];
+	s = s[2:];
+
+	switch c {
+	case:
+		return;
+
+	case 'a':  r = '\a';
+	case 'b':  r = '\b';
+	case 'f':  r = '\f';
+	case 'n':  r = '\n';
+	case 'r':  r = '\r';
+	case 't':  r = '\t';
+	case 'v':  r = '\v';
+	case '\\': r = '\\';
+
+	case '"':  r = '"';
+	case '\'': r = '\'';
+
+	case '0'..'7':
+		v := int(c-'0');
+		if len(s) < 2 {
+			return;
+		}
+		for i in 0..<len(s) {
+			d := int(s[i]-'0');
+			if d < 0 || d > 7 {
+				return;
+			}
+			v = (v<<3) | d;
+		}
+		s = s[2:];
+		if v > 0xff {
+			return;
+		}
+		r = rune(v);
+
+	case 'x', 'u', 'U':
+		count: int;
+		switch c {
+		case 'x': count = 2;
+		case 'u': count = 4;
+		case 'U': count = 8;
+		}
+
+		if len(s) < count {
+			return;
+		}
+
+		for i in 0..<count {
+			d := hex_to_int(s[i]);
+			if d < 0 {
+				return;
+			}
+			r = (r<<4) | rune(d);
+		}
+		s = s[count:];
+		if c == 'x' {
+			break;
+		}
+		if r > utf8.MAX_RUNE {
+			return;
+		}
+		multiple_bytes = true;
+	}
+
+	success = true;
+	tail_string = s;
+	return;
+}
+
+
+unquote_string :: proc(p: ^Parser, t: Token) -> (string, bool) {
+	if t.kind != Kind.String {
+		return t.lit, true;
+	}
+	s := t.lit;
+	quote := '"';
+
+	if s == `""` {
+        return "", true;
+    }
+
+	if strings.contains_rune(s, '\n') >= 0 {
+		return s, false;
+	}
+
+	if strings.contains_rune(s, '\\') < 0 && strings.contains_rune(s, quote) < 0 {
+		if quote == '"' {
+			return s, true;
+		}
+	}
+
+
+	buf_len := 3*len(s) / 2;
+	buf := make([]byte, buf_len);
+	offset := 0;
+	for len(s) > 0 {
+		r, multiple_bytes, tail_string, ok := unquote_char(s, byte(quote));
+		if !ok {
+			delete(buf);
+			return s, false;
+		}
+		s = tail_string;
+		if r < 0x80 || !multiple_bytes {
+			buf[offset] = byte(r);
+			offset += 1;
+		} else {
+			b, w := utf8.encode_rune(r);
+			copy(buf[offset:], b[:w]);
+			offset += w;
+		}
+	}
+
+	new_string := string(buf[:offset]);
+
+	append(&p.allocated_strings, new_string);
+
+	return new_string, true;
+}
+
+
+allow_token :: proc(p: ^Parser, kind: Kind) -> bool {
+	if p.curr_token.kind == kind {
+		next_token(p);
+		return true;
+	}
+	return false;
+}
+
+expect_token :: proc(p: ^Parser, kind: Kind) -> Token {
+	prev := p.curr_token;
+	if prev.kind != kind {
+		got := prev.lit;
+		if got == "\n" do got = ";";
+		error(p, prev.pos, "Expected %s, got %s", kind_to_string[kind], got);
+	}
+	next_token(p);
+	return prev;
+}
+
+expect_operator :: proc(p: ^Parser) -> Token {
+	prev := p.curr_token;
+	if !is_operator(prev.kind) {
+		error(p, prev.pos, "Expected an operator, got %s", prev.lit);
+	}
+
+
+	next_token(p);
+	return prev;
+}
+
+fix_advance :: proc(p: ^Parser) {
+	for {
+		switch t := p.curr_token; t.kind {
+		case Kind.EOF, Kind.Semicolon:
+			return;
+		}
+		next_token(p);
+	}
+}
+
+copy_value :: proc(value: Value) -> Value {
+	switch v in value {
+	case Array:
+		a := make(Array, len(v));
+		for elem, idx in v {
+			a[idx] = copy_value(elem);
+		}
+		return a;
+	case Dict:
+		d := make(Dict, cap(v));
+		for key, val in v {
+			d[key] = copy_value(val);
+		}
+		return d;
+	}
+	return value;
+}
+
+lookup_value :: proc(p: ^Parser, name: string) -> (Value, bool) {
+	for i := len(p.dict_stack)-1; i >= 0; i -= 1 {
+		d := p.dict_stack[i];
+		if val, ok := d[name]; ok {
+			return copy_value(val), true;
+		}
+	}
+
+	return nil, false;
+}
+
+parse_operand :: proc(p: ^Parser) -> (Value, Pos) {
+	tok := p.curr_token;
+	switch p.curr_token.kind {
+	case Kind.Ident:
+		next_token(p);
+		v, ok := lookup_value(p, tok.lit);
+		if !ok do error(p, tok.pos, "Undeclared identifier %s", tok.lit);
+		return v, tok.pos;
+
+	case Kind.True:
+		next_token(p);
+		return true, tok.pos;
+	case Kind.False:
+		next_token(p);
+		return false, tok.pos;
+
+	case Kind.Nil:
+		next_token(p);
+		return Nil_Value{}, tok.pos;
+
+	case Kind.Integer:
+		next_token(p);
+		return strconv.parse_i64(tok.lit), tok.pos;
+
+	case Kind.Float:
+		next_token(p);
+		return strconv.parse_f64(tok.lit), tok.pos;
+
+	case Kind.String:
+		next_token(p);
+		str, ok := unquote_string(p, tok);
+		if !ok do error(p, tok.pos, "Unable to unquote string");
+		return string(str), tok.pos;
+
+	case Kind.Open_Paren:
+		expect_token(p, Kind.Open_Paren);
+		expr, _ := parse_expr(p);
+		expect_token(p, Kind.Close_Paren);
+		return expr, tok.pos;
+
+	case Kind.Open_Bracket:
+		expect_token(p, Kind.Open_Bracket);
+		elems := make([dynamic]Value, 0, 4);
+		for p.curr_token.kind != Kind.Close_Bracket &&
+		    p.curr_token.kind != Kind.EOF {
+			elem, _ := parse_expr(p);
+			append(&elems, elem);
+
+			if p.curr_token.kind == Kind.Semicolon && p.curr_token.lit == "\n" {
+				next_token(p);
+			} else if !allow_token(p, Kind.Comma) {
+				break;
+			}
+
+		}
+		expect_token(p, Kind.Close_Bracket);
+		return Array(elems[:]), tok.pos;
+
+	case Kind.Open_Brace:
+		expect_token(p, Kind.Open_Brace);
+
+    	dict := Dict{};
+    	append(&p.dict_stack, &dict);
+    	defer pop(&p.dict_stack);
+
+		for p.curr_token.kind != Kind.Close_Brace &&
+		    p.curr_token.kind != Kind.EOF {
+		    name_tok := p.curr_token;
+		    if !allow_token(p, Kind.Ident) && !allow_token(p, Kind.String) {
+		    	name_tok = expect_token(p, Kind.Ident);
+		    }
+
+			name, ok := unquote_string(p, name_tok);
+			if !ok do error(p, tok.pos, "Unable to unquote string");
+		    expect_token(p, Kind.Assign);
+			elem, _ := parse_expr(p);
+
+			if _, ok2 := dict[name]; ok2 {
+				error(p, name_tok.pos, "Previous declaration of %s in this scope", name);
+			} else {
+				dict[name] = elem;
+			}
+
+			if p.curr_token.kind == Kind.Semicolon && p.curr_token.lit == "\n" {
+				next_token(p);
+			} else if !allow_token(p, Kind.Comma) {
+				break;
+			}
+		}
+		expect_token(p, Kind.Close_Brace);
+		return dict, tok.pos;
+
+	}
+	return nil, tok.pos;
+}
+
+parse_atom_expr :: proc(p: ^Parser, operand: Value, pos: Pos) -> (Value, Pos) {
+	loop := true;
+	for loop {
+		switch p.curr_token.kind {
+		case Kind.Period:
+			next_token(p);
+			tok := next_token(p);
+
+			switch tok.kind {
+			case Kind.Ident:
+				d, ok := operand.(Dict);
+				if !ok || d == nil {
+					error(p, tok.pos, "Expected a dictionary");
+					operand = nil;
+					continue;
+				}
+				name, usok := unquote_string(p, tok);
+				if !usok do error(p, tok.pos, "Unable to unquote string");
+				val, found := d[name];
+				if !found {
+					error(p, tok.pos, "Field %s not found in dictionary", name);
+					operand = nil;
+					continue;
+				}
+				operand = val;
+			case:
+				error(p, tok.pos, "Expected a selector, got %s", tok.kind);
+				operand = nil;
+			}
+
+		case Kind.Open_Bracket:
+			expect_token(p, Kind.Open_Bracket);
+			index, index_pos := parse_expr(p);
+			expect_token(p, Kind.Close_Bracket);
+
+
+			switch a in operand {
+			case Array:
+				i, ok := index.(i64);
+				if !ok {
+					error(p, index_pos, "Index must be an integer for an array");
+					operand = nil;
+					continue;
+				}
+
+				if 0 <= i && i < i64(len(a)) {
+					operand = a[i];
+				} else {
+					error(p, index_pos, "Index %d out of bounds range 0..%d", i, len(a));
+					operand = nil;
+					continue;
+				}
+
+			case Dict:
+				key, ok := index.(string);
+				if !ok {
+					error(p, index_pos, "Index must be a string for a dictionary");
+					operand = nil;
+					continue;
+				}
+
+				val, found := a[key];
+				if found {
+					operand = val;
+				} else {
+					error(p, index_pos, "`%s` was not found in the dictionary", key);
+					operand = nil;
+					continue;
+				}
+
+
+
+			case:
+				error(p, index_pos, "Indexing is only allowed on an array or dictionary");
+			}
+
+		case:
+			loop = false;
+		}
+	}
+
+	return operand, pos;
+}
+
+parse_unary_expr :: proc(p: ^Parser) -> (Value, Pos) {
+	op := p.curr_token;
+	switch p.curr_token.kind {
+	case Kind.At:
+		next_token(p);
+		tok := expect_token(p, Kind.String);
+		v, ok := lookup_value(p, tok.lit);
+		if !ok do error(p, tok.pos, "Undeclared identifier %s", tok.lit);
+		return parse_atom_expr(p, v, tok.pos);
+
+	case Kind.Add, Kind.Sub:
+		next_token(p);
+		// TODO(bill): Calcuate values as you go!
+		expr, pos := parse_unary_expr(p);
+
+		switch e in expr {
+		case i64: if op.kind == Kind.Sub do return -e, pos;
+		case f64: if op.kind == Kind.Sub do return -e, pos;
+		case:
+			error(p, op.pos, "Unary operator %s can only be used on integers or floats", op.lit);
+			return nil, op.pos;
+		}
+
+		return expr, op.pos;
+
+	case Kind.Not:
+		next_token(p);
+		expr, _ := parse_unary_expr(p);
+		if v, ok := expr.(bool); ok {
+			return !v, op.pos;
+		}
+		error(p, op.pos, "Unary operator %s can only be used on booleans", op.lit);
+		return nil, op.pos;
+	}
+
+	return parse_atom_expr(p, parse_operand(p));
+}
+
+
+value_order :: proc(v: Value) -> int {
+	switch _ in v {
+	case bool, string:
+		return 1;
+	case i64:
+		return 2;
+	case f64:
+		return 3;
+	}
+	return 0;
+}
+
+match_values :: proc(left, right: ^Value) -> bool {
+	if value_order(right^) < value_order(left^) {
+		return match_values(right, left);
+	}
+
+	switch x in left^ {
+	case:
+		right^ = left^;
+	case bool, string:
+		return true;
+	case i64:
+		switch y in right^ {
+		case i64:
+			return true;
+		case f64:
+			left^ = f64(x);
+			return true;
+		}
+
+	case f64:
+		switch y in right {
+		case f64:
+			return true;
+		}
+	}
+
+	return false;
+}
+
+calculate_binary_value :: proc(p: ^Parser, op: Kind, x, y: Value) -> (Value, bool) {
+	// TODO(bill): Calculate value as you go!
+	match_values(&x, &y);
+
+
+	switch a in x {
+	case: return x, true;
+
+	case bool:
+		b, ok := y.(bool);
+		if !ok do return nil, false;
+		switch op {
+		case Kind.Eq:    return a == b, true;
+		case Kind.NotEq: return a != b, true;
+		case Kind.And:   return a && b, true;
+		case Kind.Or:    return a || b, true;
+		}
+
+	case i64:
+		b, ok := y.(i64);
+		if !ok do return nil, false;
+		switch op {
+		case Kind.Add:   return a + b, true;
+		case Kind.Sub:   return a - b, true;
+		case Kind.Mul:   return a * b, true;
+		case Kind.Quo:   return a / b, true;
+		case Kind.Rem:   return a % b, true;
+		case Kind.Eq:    return a == b, true;
+		case Kind.NotEq: return a != b, true;
+		case Kind.Lt:    return a <  b, true;
+		case Kind.Gt:    return a >  b, true;
+		case Kind.LtEq:  return a <= b, true;
+		case Kind.GtEq:  return a >= b, true;
+		}
+
+	case f64:
+		b, ok := y.(f64);
+		if !ok do return nil, false;
+
+		switch op {
+		case Kind.Add:   return a + b, true;
+		case Kind.Sub:   return a - b, true;
+		case Kind.Mul:   return a * b, true;
+		case Kind.Quo:   return a / b, true;
+		case Kind.Eq:    return a == b, true;
+		case Kind.NotEq: return a != b, true;
+		case Kind.Lt:    return a <  b, true;
+		case Kind.Gt:    return a >  b, true;
+		case Kind.LtEq:  return a <= b, true;
+		case Kind.GtEq:  return a >= b, true;
+		}
+
+	case string:
+		b, ok := y.(string);
+		if !ok do return nil, false;
+
+		switch op {
+		case Kind.Add:
+			n := len(a) + len(b);
+			data := make([]byte, n);
+			copy(data[:], cast([]byte)a);
+			copy(data[len(a):], cast([]byte)b);
+			s := string(data);
+			append(&p.allocated_strings, s);
+			return s, true;
+
+		case Kind.Eq:    return a == b, true;
+		case Kind.NotEq: return a != b, true;
+		case Kind.Lt:    return a <  b, true;
+		case Kind.Gt:    return a >  b, true;
+		case Kind.LtEq:  return a <= b, true;
+		case Kind.GtEq:  return a >= b, true;
+		}
+	}
+
+	return nil, false;
+}
+
+parse_binary_expr :: proc(p: ^Parser, prec_in: int) -> (Value, Pos) {
+	expr, pos := parse_unary_expr(p);
+	for prec := precedence(p.curr_token.kind); prec >= prec_in; prec -= 1 {
+		for {
+			op := p.curr_token;
+			op_prec := precedence(op.kind);
+			if op_prec != prec {
+				break;
+			}
+			expect_operator(p);
+
+			if op.kind == Kind.Question {
+				cond := expr;
+				x, _ := parse_expr(p);
+				expect_token(p, Kind.Colon);
+				y, _ := parse_expr(p);
+
+				if t, ok := cond.(bool); ok {
+					expr = t ? x : y;
+				} else {
+					error(p, pos, "Condition must be a boolean");
+				}
+
+			} else {
+				right, right_pos := parse_binary_expr(p, prec+1);
+				if right == nil {
+					error(p, right_pos, "Expected expression on the right-hand side of the binary operator %s", op.lit);
+				}
+				left := expr;
+				ok: bool;
+				expr, ok = calculate_binary_value(p, op.kind, left, right);
+				if !ok {
+					error(p, pos, "Invalid binary operation");
+				}
+			}
+		}
+	}
+	return expr, pos;
+}
+
+parse_expr :: proc(p: ^Parser) -> (Value, Pos) {
+	return parse_binary_expr(p, 1);
+}
+
+expect_semicolon :: proc(p: ^Parser) {
+	kind := p.curr_token.kind;
+
+	switch kind {
+	case Kind.Comma:
+		error(p, p.curr_token.pos, "Expected ';', got ','");
+		next_token(p);
+	case Kind.Semicolon:
+		next_token(p);
+	case Kind.EOF:
+		// okay
+	case:
+		error(p, p.curr_token.pos, "Expected ';', got %s", p.curr_token.lit);
+		fix_advance(p);
+	}
+}
+
+parse_assignment :: proc(p: ^Parser) -> bool {
+	top_dict :: proc(p: ^Parser) -> ^Dict {
+		assert(len(p.dict_stack) > 0);
+		return p.dict_stack[len(p.dict_stack)-1];
+	}
+
+	if p.curr_token.kind == Kind.Semicolon {
+		next_token(p);
+		return true;
+	}
+	if p.curr_token.kind == Kind.EOF {
+		return false;
+	}
+
+	tok := p.curr_token;
+	if allow_token(p, Kind.Ident) || allow_token(p, Kind.String) {
+		expect_token(p, Kind.Assign);
+		name, ok := unquote_string(p, tok);
+		if !ok do error(p, tok.pos, "Unable to unquote string");
+		expr, _ := parse_expr(p);
+		d := top_dict(p);
+		if _, ok2 := d[name]; ok2 {
+			error(p, tok.pos, "Previous declaration of %s", name);
+		} else {
+			d[name] = expr;
+		}
+		expect_semicolon(p);
+		return true;
+	}
+	error(p, tok.pos, "Expected an assignment, got %s", kind_to_string[tok.kind]);
+	fix_advance(p);
+	return false;
+}

core/encoding/cel/doc.odin

@@ -0,0 +1,51 @@
+/*
+	package cel
+
+	sample := `
+	x = 123;
+	y = 321.456;
+	z = x * (y - 1) / 2;
+	w = "foo" + "bar";
+
+	# This is a comment
+
+	asd = "Semicolons are optional"
+
+	a = {id = {b = 123}} # Dict
+	b = a.id.b
+
+	f = [1, 4, 9] # Array
+	g = f[2]
+
+	h = x < y and w == "foobar"
+	i = h ? 123 : "google"
+
+	j = nil
+
+	"127.0.0.1" = "value" # Keys can be strings
+
+	"foo" = {
+		"bar" = {
+			"baz" = 123, # optional commas if newline is present
+			"zab" = 456,
+			"abz" = 789,
+		},
+	};
+
+	bar = @"foo"["bar"].baz
+	`;
+
+
+	main :: proc() {
+		p, ok := create_from_string(sample);
+		if !ok {
+			return;
+		}
+		defer destroy(p);
+
+		if p.error_count == 0 {
+			print(p);
+		}
+	}
+*/
+package cel

core/encoding/cel/token.odin

@@ -0,0 +1,520 @@
+package cel
+
+import "core:fmt"
+import "core:unicode/utf8"
+
+using Kind :: enum {
+	Illegal,
+	EOF,
+	Comment,
+
+	_literal_start,
+		Ident,
+		Integer,
+		Float,
+		Char,
+		String,
+	_literal_end,
+
+	_keyword_start,
+		True,  // true
+		False, // false
+		Nil,   // nil
+	_keyword_end,
+
+
+	_operator_start,
+		Question, // ?
+
+		And,   // and
+		Or,    // or
+
+		Add,   // +
+		Sub,   // -
+		Mul,   // *
+		Quo,   // /
+		Rem,   // %
+
+		Not,   // !
+
+		Eq,    // ==
+		NotEq, // !=
+		Lt,    // <
+		Gt,    // >
+		LtEq,  // <=
+		GtEq,  // >=
+
+		At,    // @
+	_operator_end,
+
+	_punc_start,
+		Assign, // =
+
+		Open_Paren,    // (
+		Close_Paren,   // )
+		Open_Bracket,  // [
+		Close_Bracket, // ]
+		Open_Brace,    // {
+		Close_Brace,   // }
+
+		Colon,     // :
+		Semicolon, // ;
+		Comma,     // ,
+		Period,    // .
+	_punc_end,
+}
+
+
+Pos :: struct {
+	file:   string,
+	line:   int,
+	column: int,
+}
+
+Token :: struct {
+	kind:      Kind,
+	using pos: Pos,
+	lit:       string,
+}
+
+Tokenizer :: struct {
+	src:         []byte,
+
+	file:        string, // May not be used
+
+	curr_rune:   rune,
+	offset:      int,
+	read_offset: int,
+	line_offset: int,
+	line_count:  int,
+
+	insert_semi: bool,
+
+	error_count: int,
+}
+
+
+keywords := map[string]Kind{
+	"true"  = True,
+	"false" = False,
+	"nil"   = Nil,
+	"and"   = And,
+	"or"    = Or,
+};
+
+kind_to_string := [len(Kind)]string{
+	"illegal",
+	"EOF",
+	"comment",
+
+	"",
+	"identifier",
+	"integer",
+	"float",
+	"character",
+	"string",
+	"",
+
+	"",
+	"true", "false", "nil",
+	"",
+
+	"",
+	"?", "and", "or",
+	"+", "-", "*", "/", "%",
+	"!",
+	"==", "!=", "<", ">", "<=", ">=",
+	"@",
+	"",
+
+	"",
+	"=",
+	"(", ")",
+	"[", "]",
+	"{", "}",
+	":", ";", ",", ".",
+	"",
+};
+
+precedence :: proc(op: Kind) -> int {
+	switch op {
+	case Question:
+		return 1;
+	case Or:
+		return 2;
+	case And:
+		return 3;
+	case Eq, NotEq, Lt, Gt, LtEq, GtEq:
+		return 4;
+	case Add, Sub:
+		return 5;
+	case Mul, Quo, Rem:
+		return 6;
+	}
+	return 0;
+}
+
+
+token_lookup :: proc(ident: string) -> Kind {
+	if tok, is_keyword := keywords[ident]; is_keyword {
+		return tok;
+	}
+	return Ident;
+}
+
+is_literal  :: proc(tok: Kind) -> bool do return _literal_start  < tok && tok < _literal_end;
+is_operator :: proc(tok: Kind) -> bool do return _operator_start < tok && tok < _operator_end;
+is_keyword  :: proc(tok: Kind) -> bool do return _keyword_start  < tok && tok < _keyword_end;
+
+
+tokenizer_init :: proc(t: ^Tokenizer, src: []byte, file := "") {
+	t.src = src;
+	t.file = file;
+	t.curr_rune   = ' ';
+	t.offset      = 0;
+	t.read_offset = 0;
+	t.line_offset = 0;
+	t.line_count  = 1;
+
+	advance_to_next_rune(t);
+	if t.curr_rune == utf8.RUNE_BOM {
+		advance_to_next_rune(t);
+	}
+}
+
+token_error :: proc(t: ^Tokenizer, msg: string, args: ..any) {
+	fmt.printf_err("%s(%d:%d) Error: ", t.file, t.line_count, t.read_offset-t.line_offset+1);
+	fmt.printf_err(msg, ..args);
+	fmt.println_err();
+	t.error_count += 1;
+}
+
+advance_to_next_rune :: proc(t: ^Tokenizer) {
+	if t.read_offset < len(t.src) {
+		t.offset = t.read_offset;
+		if t.curr_rune == '\n' {
+			t.line_offset = t.offset;
+			t.line_count += 1;
+		}
+		r, w := rune(t.src[t.read_offset]), 1;
+		switch {
+		case r == 0:
+			token_error(t, "Illegal character NUL");
+		case r >= utf8.RUNE_SELF:
+			r, w = utf8.decode_rune(t.src[t.read_offset:]);
+			if r == utf8.RUNE_ERROR && w == 1 {
+				token_error(t, "Illegal utf-8 encoding");
+			} else if r == utf8.RUNE_BOM && t.offset > 0 {
+				token_error(t, "Illegal byte order mark");
+			}
+		}
+
+		t.read_offset += w;
+		t.curr_rune = r;
+	} else {
+		t.offset = len(t.src);
+		if t.curr_rune == '\n' {
+			t.line_offset = t.offset;
+			t.line_count += 1;
+		}
+		t.curr_rune = utf8.RUNE_EOF;
+	}
+}
+
+
+get_pos :: proc(t: ^Tokenizer) -> Pos {
+	return Pos {
+		file   = t.file,
+		line   = t.line_count,
+		column = t.offset - t.line_offset + 1,
+	};
+}
+
+is_letter :: proc(r: rune) -> bool {
+	switch r {
+	case 'a'..'z', 'A'..'Z', '_':
+		return true;
+	}
+	return false;
+}
+
+is_digit :: proc(r: rune) -> bool {
+	switch r {
+	case '0'..'9':
+		return true;
+	}
+	return false;
+}
+
+skip_whitespace :: proc(t: ^Tokenizer) {
+	loop: for {
+		switch t.curr_rune {
+		case '\n':
+			if t.insert_semi {
+				break loop;
+			}
+			fallthrough;
+		case ' ', '\t', '\r', '\v', '\f':
+			advance_to_next_rune(t);
+
+		case:
+			break loop;
+		}
+	}
+}
+
+scan_identifier :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset;
+	for is_letter(t.curr_rune) || is_digit(t.curr_rune) {
+		advance_to_next_rune(t);
+	}
+	return string(t.src[offset : t.offset]);
+}
+
+digit_value :: proc(r: rune) -> int {
+	switch r {
+	case '0'..'9': return int(r - '0');
+	case 'a'..'f': return int(r - 'a' + 10);
+	case 'A'..'F': return int(r - 'A' + 10);
+	}
+	return 16;
+}
+
+scan_number :: proc(t: ^Tokenizer, seen_decimal_point: bool) -> (Kind, string) {
+	scan_mantissa :: proc(t: ^Tokenizer, base: int) {
+		for digit_value(t.curr_rune) < base || t.curr_rune == '_' {
+			advance_to_next_rune(t);
+		}
+	}
+	scan_exponent :: proc(t: ^Tokenizer, tok: Kind, offset: int) -> (Kind, string) {
+		if t.curr_rune == 'e' || t.curr_rune == 'E' {
+			tok = Float;
+			advance_to_next_rune(t);
+			if t.curr_rune == '-' || t.curr_rune == '+' {
+				advance_to_next_rune(t);
+			}
+			if digit_value(t.curr_rune) < 10 {
+				scan_mantissa(t, 10);
+			} else {
+				token_error(t, "Illegal floating point exponent");
+			}
+		}
+		return tok, string(t.src[offset : t.offset]);
+	}
+	scan_fraction :: proc(t: ^Tokenizer, tok: Kind, offset: int) -> (Kind, string) {
+		if t.curr_rune == '.' {
+			tok = Float;
+			advance_to_next_rune(t);
+			scan_mantissa(t, 10);
+		}
+
+		return scan_exponent(t, tok, offset);
+	}
+
+	offset := t.offset;
+	tok := Integer;
+
+	if seen_decimal_point {
+		offset -= 1;
+		tok = Float;
+		scan_mantissa(t, 10);
+		return scan_exponent(t, tok, offset);
+	}
+
+	if t.curr_rune == '0' {
+		offset = t.offset;
+		advance_to_next_rune(t);
+		switch t.curr_rune {
+		case 'b', 'B':
+			advance_to_next_rune(t);
+			scan_mantissa(t, 2);
+			if t.offset - offset <= 2 {
+				token_error(t, "Illegal binary number");
+			}
+		case 'o', 'O':
+			advance_to_next_rune(t);
+			scan_mantissa(t, 8);
+			if t.offset - offset <= 2 {
+				token_error(t, "Illegal octal number");
+			}
+		case 'x', 'X':
+			advance_to_next_rune(t);
+			scan_mantissa(t, 16);
+			if t.offset - offset <= 2 {
+				token_error(t, "Illegal hexadecimal number");
+			}
+		case:
+			scan_mantissa(t, 10);
+			switch t.curr_rune {
+			case '.', 'e', 'E':
+				return scan_fraction(t, tok, offset);
+			}
+		}
+
+		return tok, string(t.src[offset:t.offset]);
+	}
+
+	scan_mantissa(t, 10);
+
+	return scan_fraction(t, tok, offset);
+}
+
+scan :: proc(t: ^Tokenizer) -> Token {
+	skip_whitespace(t);
+
+	offset := t.offset;
+
+	tok: Kind;
+	pos := get_pos(t);
+	lit: string;
+
+	insert_semi := false;
+
+
+	switch r := t.curr_rune; {
+	case is_letter(r):
+		insert_semi = true;
+		lit = scan_identifier(t);
+		tok = Ident;
+		if len(lit) > 1 {
+			tok = token_lookup(lit);
+		}
+
+	case '0' <= r && r <= '9':
+		insert_semi = true;
+		tok, lit = scan_number(t, false);
+
+	case:
+		advance_to_next_rune(t);
+		switch r {
+		case -1:
+			if t.insert_semi {
+				t.insert_semi = false;
+				return Token{Semicolon, pos, "\n"};
+			}
+			return Token{EOF, pos, "\n"};
+
+		case '\n':
+			t.insert_semi = false;
+			return Token{Semicolon, pos, "\n"};
+
+		case '"':
+			insert_semi = true;
+			quote := r;
+			tok = String;
+			for {
+				this_r := t.curr_rune;
+				if this_r == '\n' || r < 0 {
+					token_error(t, "String literal not terminated");
+					break;
+				}
+				advance_to_next_rune(t);
+				if this_r == quote {
+					break;
+				}
+				// TODO(bill); Handle properly
+				if this_r == '\\' && t.curr_rune == quote {
+					advance_to_next_rune(t);
+				}
+			}
+
+			lit = string(t.src[offset+1:t.offset-1]);
+
+
+		case '#':
+			for t.curr_rune != '\n' && t.curr_rune >= 0 {
+				advance_to_next_rune(t);
+			}
+			if t.insert_semi {
+				t.insert_semi = false;
+				return Token{Semicolon, pos, "\n"};
+			}
+			// Recursive!
+			return scan(t);
+
+		case '?': tok = Question;
+		case ':': tok = Colon;
+		case '@': tok = At;
+
+		case ';':
+			tok = Semicolon;
+			lit = ";";
+		case ',': tok = Comma;
+
+		case '(':
+			tok = Open_Paren;
+		case ')':
+			insert_semi = true;
+			tok = Close_Paren;
+
+		case '[':
+			tok = Open_Bracket;
+		case ']':
+			insert_semi = true;
+			tok = Close_Bracket;
+
+		case '{':
+			tok = Open_Brace;
+		case '}':
+			insert_semi = true;
+			tok = Close_Brace;
+
+		case '+': tok = Add;
+		case '-': tok = Sub;
+		case '*': tok = Mul;
+		case '/': tok = Quo;
+		case '%': tok = Rem;
+
+		case '!':
+			tok = Not;
+			if t.curr_rune == '=' {
+				advance_to_next_rune(t);
+				tok = NotEq;
+			}
+
+		case '=':
+			tok = Assign;
+			if t.curr_rune == '=' {
+				advance_to_next_rune(t);
+				tok = Eq;
+			}
+
+		case '<':
+			tok = Lt;
+			if t.curr_rune == '=' {
+				advance_to_next_rune(t);
+				tok = LtEq;
+			}
+
+		case '>':
+			tok = Gt;
+			if t.curr_rune == '=' {
+				advance_to_next_rune(t);
+				tok = GtEq;
+			}
+
+		case '.':
+			if '0' <= t.curr_rune && t.curr_rune <= '9' {
+				insert_semi = true;
+				tok, lit = scan_number(t, true);
+			} else {
+				tok = Period;
+			}
+
+		case:
+			if r != utf8.RUNE_BOM {
+				token_error(t, "Illegal character '%r'", r);
+			}
+			insert_semi = t.insert_semi;
+			tok = Illegal;
+		}
+	}
+
+	t.insert_semi = insert_semi;
+
+	if lit == "" {
+		lit = string(t.src[offset:t.offset]);
+	}
+
+	return Token{tok, pos, lit};
+}

core/encoding/json/marshal.odin

@@ -0,0 +1,331 @@
+package json
+
+import "core:mem"
+import "core:math/bits"
+import "core:runtime"
+import "core:strconv"
+import "core:strings"
+import "core:types"
+
+Marshal_Error :: enum {
+	None,
+	Unsupported_Type,
+}
+
+marshal :: proc(v: any, allocator := context.allocator) -> ([]byte, Marshal_Error) {
+	b := strings.make_builder(allocator);
+
+	err := marshal_arg(&b, v);
+
+	if err != Marshal_Error.None {
+		strings.destroy_builder(&b);
+		return nil, err;
+	}
+	if len(b.buf) == 0 {
+		strings.destroy_builder(&b);
+		return nil, err;
+	}
+	return b.buf[:], err;
+}
+
+
+marshal_arg :: proc(b: ^strings.Builder, v: any) -> Marshal_Error {
+	using strings;
+	using runtime;
+	if v == nil {
+		write_string(b, "null");
+		return Marshal_Error.None;
+	}
+
+	ti := type_info_base(type_info_of(v.id));
+	a := any{v.data, ti.id};
+
+	switch info in ti.variant {
+	case Type_Info_Named:
+		panic("Unreachable");
+
+	case Type_Info_Integer:
+		buf: [21]byte;
+		u: u64;
+		switch i in a {
+		case i8:      u = u64(i);
+		case i16:     u = u64(i);
+		case i32:     u = u64(i);
+		case i64:     u = u64(i);
+		case int:     u = u64(i);
+		case u8:      u = u64(i);
+		case u16:     u = u64(i);
+		case u32:     u = u64(i);
+		case u64:     u = u64(i);
+		case uint:    u = u64(i);
+		case uintptr: u = u64(i);
+
+		case i16le: u = u64(i);
+		case i32le: u = u64(i);
+		case i64le: u = u64(i);
+		case u16le: u = u64(i);
+		case u32le: u = u64(i);
+		case u64le: u = u64(i);
+
+		case i16be: u = u64(i);
+		case i32be: u = u64(i);
+		case i64be: u = u64(i);
+		case u16be: u = u64(i);
+		case u32be: u = u64(i);
+		case u64be: u = u64(i);
+		}
+
+		s := strconv.append_bits(buf[:], u, 10, info.signed, 8*ti.size, "0123456789", nil);
+		write_string(b, s);
+
+
+	case Type_Info_Rune:
+		r := a.(rune);
+		write_byte(b, '"');
+		write_escaped_rune(b, r, '"', true);
+		write_byte(b, '"');
+
+	case Type_Info_Float:
+		val: f64;
+		switch f in a {
+		case f32: val = f64(f);
+		case f64: val = f64(f);
+		}
+
+		buf: [386]byte;
+
+		str := strconv.append_float(buf[1:], val, 'f', 2*ti.size, 8*ti.size);
+		str = string(buf[:len(str)+1]);
+		if str[1] == '+' || str[1] == '-' {
+			str = str[1:];
+		} else {
+			str[0] = '+';
+		}
+		if str[0] == '+' {
+			str = str[1:];
+		}
+
+		write_string(b, str);
+
+	case Type_Info_Complex:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_String:
+		switch s in a {
+		case string:  write_quoted_string(b, s);
+		case cstring: write_quoted_string(b, string(s));
+		}
+
+	case Type_Info_Boolean:
+		val: bool;
+		switch b in a {
+		case bool: val = bool(b);
+		case b8:   val = bool(b);
+		case b16:  val = bool(b);
+		case b32:  val = bool(b);
+		case b64:  val = bool(b);
+		}
+		write_string(b, val ? "true" : "false");
+
+	case Type_Info_Any:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Type_Id:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Pointer:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Procedure:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Tuple:
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Array:
+		write_byte(b, '[');
+		for i in 0..<info.count {
+			if i > 0 do write_string(b, ", ");
+
+			data := uintptr(v.data) + uintptr(i*info.elem_size);
+			marshal_arg(b, any{rawptr(data), info.elem.id});
+		}
+		write_byte(b, ']');
+
+	case Type_Info_Dynamic_Array:
+		write_byte(b, '[');
+		array := cast(^mem.Raw_Dynamic_Array)v.data;
+		for i in 0..<array.len {
+			if i > 0 do write_string(b, ", ");
+
+			data := uintptr(array.data) + uintptr(i*info.elem_size);
+			marshal_arg(b, any{rawptr(data), info.elem.id});
+		}
+		write_byte(b, ']');
+
+	case Type_Info_Slice:
+		write_byte(b, '[');
+		slice := cast(^mem.Raw_Slice)v.data;
+		for i in 0..<slice.len {
+			if i > 0 do write_string(b, ", ");
+
+			data := uintptr(slice.data) + uintptr(i*info.elem_size);
+			marshal_arg(b, any{rawptr(data), info.elem.id});
+		}
+		write_byte(b, ']');
+
+	case Type_Info_Map:
+		m := (^mem.Raw_Map)(v.data);
+
+		write_byte(b, '{');
+		if m != nil {
+			if info.generated_struct == nil {
+				return Marshal_Error.Unsupported_Type;
+			}
+			entries    := &m.entries;
+			gs         := type_info_base(info.generated_struct).variant.(Type_Info_Struct);
+			ed         := type_info_base(gs.types[1]).variant.(Type_Info_Dynamic_Array);
+			entry_type := ed.elem.variant.(Type_Info_Struct);
+			entry_size := ed.elem_size;
+
+			for i in 0..<entries.len {
+				if i > 0 do write_string(b, ", ");
+
+				data := uintptr(entries.data) + uintptr(i*entry_size);
+				header := cast(^Map_Entry_Header)data;
+
+				if types.is_string(info.key) {
+					marshal_arg(b, header.key.str);
+				} else {
+					marshal_arg(b, any{rawptr(&header.key.hash), info.key.id});
+				}
+
+				write_string(b, ": ");
+
+				value := data + entry_type.offsets[2];
+				marshal_arg(b, any{rawptr(value), info.value.id});
+			}
+		}
+		write_byte(b, '}');
+
+	case Type_Info_Struct:
+		write_byte(b, '{');
+		for name, i in info.names {
+			if i > 0 do write_string(b, ", ");
+			write_quoted_string(b, name);
+			write_string(b, ": ");
+
+			id := info.types[i].id;
+			data := rawptr(uintptr(v.data) + info.offsets[i]);
+			marshal_arg(b, any{data, id});
+		}
+		write_byte(b, '}');
+
+	case Type_Info_Union:
+		tag_ptr := uintptr(v.data) + info.tag_offset;
+		tag_any := any{rawptr(tag_ptr), info.tag_type.id};
+
+		tag: i64 = -1;
+		switch i in tag_any {
+		case u8:   tag = i64(i);
+		case i8:   tag = i64(i);
+		case u16:  tag = i64(i);
+		case i16:  tag = i64(i);
+		case u32:  tag = i64(i);
+		case i32:  tag = i64(i);
+		case u64:  tag = i64(i);
+		case i64:  tag = i64(i);
+		case: panic("Invalid union tag type");
+		}
+
+		if v.data == nil || tag == 0 {
+			write_string(b, "null");
+		} else {
+			id := info.variants[tag-1].id;
+			marshal_arg(b, any{v.data, id});
+		}
+
+	case Type_Info_Enum:
+		return marshal_arg(b, any{v.data, info.base.id});
+
+	case Type_Info_Bit_Field:
+		data: u64 = 0;
+		switch ti.size {
+		case 1: data = cast(u64) (^u8)(v.data)^;
+		case 2: data = cast(u64)(^u16)(v.data)^;
+		case 4: data = cast(u64)(^u32)(v.data)^;
+		case 8: data = cast(u64)(^u64)(v.data)^;
+		}
+
+		write_byte(b, '{');
+		for name, i in info.names {
+			if i > 0 do write_string(b, ", ");
+
+			bits := u64(info.bits[i]);
+			offset := u64(info.offsets[i]);
+			marshal_arg(b, name);
+			write_string(b, ": ");
+
+			n := 8*u64(size_of(u64));
+			sa := n - bits;
+			u := data>>offset;
+			u <<= sa;
+			u >>= sa;
+
+			write_u64(b, u, 10);
+		}
+		write_byte(b, '}');
+
+	case Type_Info_Bit_Set:
+		is_bit_set_different_endian_to_platform :: proc(ti: ^runtime.Type_Info) -> bool {
+			if ti == nil {
+				return false;
+			}
+			ti = runtime.type_info_base(ti);
+			switch info in ti.variant {
+			case runtime.Type_Info_Integer:
+				using runtime.Type_Info_Endianness;
+				switch info.endianness {
+				case Platform: return false;
+				case Little:   return ODIN_ENDIAN != "little";
+				case Big:      return ODIN_ENDIAN != "big";
+				}
+			}
+			return false;
+		}
+
+		bit_data: u64;
+		bit_size := u64(8*ti.size);
+
+		do_byte_swap := is_bit_set_different_endian_to_platform(info.underlying);
+
+		switch bit_size {
+		case  0: bit_data = 0;
+		case  8:
+			x := (^u8)(v.data)^;
+			bit_data = u64(x);
+		case 16:
+			x := (^u16)(v.data)^;
+			if do_byte_swap do x = bits.byte_swap(x);
+			bit_data = u64(x);
+		case 32:
+			x := (^u32)(v.data)^;
+			if do_byte_swap do x = bits.byte_swap(x);
+			bit_data = u64(x);
+		case 64:
+			x := (^u64)(v.data)^;
+			if do_byte_swap do x = bits.byte_swap(x);
+			bit_data = u64(x);
+		case: panic("unknown bit_size size");
+		}
+		write_u64(b, bit_data);
+
+
+		return Marshal_Error.Unsupported_Type;
+
+	case Type_Info_Opaque:
+		return Marshal_Error.Unsupported_Type;
+	}
+
+	return Marshal_Error.None;
+}

core/encoding/json/parser.odin

@@ -0,0 +1,455 @@
+package json
+
+import "core:mem"
+import "core:unicode/utf8"
+import "core:strconv"
+
+Parser :: struct {
+	tok:        Tokenizer,
+	prev_token: Token,
+	curr_token: Token,
+	spec:       Specification,
+	allocator:  mem.Allocator,
+}
+
+make_parser :: proc(data: []byte, spec := Specification.JSON, allocator := context.allocator) -> Parser {
+	p: Parser;
+	p.tok = make_tokenizer(data, spec);
+	p.spec = spec;
+	p.allocator = allocator;
+	assert(p.allocator.procedure != nil);
+	advance_token(&p);
+	return p;
+}
+
+parse :: proc(data: []byte, spec := Specification.JSON, allocator := context.allocator) -> (Value, Error) {
+	context.allocator = allocator;
+	p := make_parser(data, spec, allocator);
+
+	if p.spec == Specification.JSON5 {
+		return parse_value(&p);
+	}
+	return parse_object(&p);
+}
+
+token_end_pos :: proc(tok: Token) -> Pos {
+	end := tok.pos;
+	end.offset += len(tok.text);
+	return end;
+}
+
+advance_token :: proc(p: ^Parser) -> (Token, Error) {
+	err: Error;
+	p.prev_token = p.curr_token;
+	p.curr_token, err = get_token(&p.tok);
+	return p.prev_token, err;
+}
+
+
+allow_token :: proc(p: ^Parser, kind: Kind) -> bool {
+	if p.curr_token.kind == kind {
+		advance_token(p);
+		return true;
+	}
+	return false;
+}
+
+expect_token :: proc(p: ^Parser, kind: Kind) -> Error {
+	prev := p.curr_token;
+	advance_token(p);
+	if prev.kind == kind {
+		return Error.None;
+	}
+	return Error.Unexpected_Token;
+}
+
+
+
+parse_value :: proc(p: ^Parser) -> (value: Value, err: Error) {
+	value.pos = p.curr_token.pos;
+	defer value.end = token_end_pos(p.prev_token);
+
+	token := p.curr_token;
+	switch token.kind {
+	case Kind.Null:
+		value.value = Null{};
+		advance_token(p);
+		return;
+	case Kind.False:
+		value.value = Boolean(false);
+		advance_token(p);
+		return;
+	case Kind.True:
+		value.value = Boolean(true);
+		advance_token(p);
+		return;
+
+	case Kind.Integer:
+		value.value = Integer(strconv.parse_i64(token.text));
+		advance_token(p);
+		return;
+	case Kind.Float:
+		value.value = Float(strconv.parse_f64(token.text));
+		advance_token(p);
+		return;
+	case Kind.String:
+		value.value = String(unquote_string(token, p.spec, p.allocator));
+		advance_token(p);
+		return;
+
+	case Kind.Open_Brace:
+		return parse_object(p);
+
+	case Kind.Open_Bracket:
+		return parse_array(p);
+
+	case:
+		if p.spec == Specification.JSON5 {
+			switch token.kind {
+			case Kind.Infinity:
+				inf: u64 = 0x7ff0000000000000;
+				if token.text[0] == '-' {
+					inf = 0xfff0000000000000;
+				}
+				value.value = transmute(f64)inf;
+				advance_token(p);
+				return;
+			case Kind.NaN:
+				nan: u64 = 0x7ff7ffffffffffff;
+				if token.text[0] == '-' {
+					nan = 0xfff7ffffffffffff;
+				}
+				value.value = transmute(f64)nan;
+				advance_token(p);
+				return;
+			}
+		}
+	}
+
+	err = Error.Unexpected_Token;
+	advance_token(p);
+	return;
+}
+
+parse_array :: proc(p: ^Parser) -> (value: Value, err: Error) {
+	value.pos = p.curr_token.pos;
+	defer value.end = token_end_pos(p.prev_token);
+	if err = expect_token(p, Kind.Open_Bracket); err != Error.None {
+		return;
+	}
+
+	array: Array;
+	array.allocator = p.allocator;
+	defer if err != Error.None {
+		for elem in array {
+			destroy_value(elem);
+		}
+		delete(array);
+	}
+
+	for p.curr_token.kind != Kind.Close_Bracket {
+		elem, elem_err := parse_value(p);
+		if elem_err != Error.None {
+			err = elem_err;
+			return;
+		}
+		append(&array, elem);
+
+		// Disallow trailing commas for the time being
+		if allow_token(p, Kind.Comma) {
+			continue;
+		} else {
+			break;
+		}
+	}
+
+	if err = expect_token(p, Kind.Close_Bracket); err != Error.None {
+		return;
+	}
+
+	value.value = array;
+	return;
+}
+
+clone_string :: proc(s: string, allocator: mem.Allocator) -> string {
+	n := len(s);
+	b := make([]byte, n+1, allocator);
+	copy(b, cast([]byte)s);
+	b[n] = 0;
+	return string(b[:n]);
+}
+
+parse_object_key :: proc(p: ^Parser) -> (key: string, err: Error) {
+	tok := p.curr_token;
+	if p.spec == Specification.JSON5 {
+		if tok.kind == Kind.String {
+			expect_token(p, Kind.String);
+			key = unquote_string(tok, p.spec, p.allocator);
+			return;
+		} else if tok.kind == Kind.Ident {
+			expect_token(p, Kind.Ident);
+			key = clone_string(tok.text, p.allocator);
+			return;
+		}
+	}
+	if tok_err := expect_token(p, Kind.String); tok_err != Error.None {
+		err = Error.Expected_String_For_Object_Key;
+		return;
+	}
+	key = unquote_string(tok, p.spec, p.allocator);
+	return;
+}
+
+parse_object :: proc(p: ^Parser) -> (value: Value, err: Error) {
+	value.pos = p.curr_token.pos;
+	defer value.end = token_end_pos(p.prev_token);
+
+	if err = expect_token(p, Kind.Open_Brace); err != Error.None {
+		value.pos = p.curr_token.pos;
+		return;
+	}
+
+	obj: Object;
+	obj.allocator = p.allocator;
+	defer if err != Error.None {
+		for key, elem in obj {
+			delete(key, p.allocator);
+			destroy_value(elem);
+		}
+		delete(obj);
+	}
+
+	for p.curr_token.kind != Kind.Close_Brace {
+		key: string;
+		key, err = parse_object_key(p);
+		if err != Error.None {
+			delete(key, p.allocator);
+			value.pos = p.curr_token.pos;
+			return;
+		}
+
+		if colon_err := expect_token(p, Kind.Colon); colon_err != Error.None {
+			err = Error.Expected_Colon_After_Key;
+			value.pos = p.curr_token.pos;
+			return;
+		}
+
+		elem, elem_err := parse_value(p);
+		if elem_err != Error.None {
+			err = elem_err;
+			value.pos = p.curr_token.pos;
+			return;
+		}
+
+		if key in obj {
+			err = Error.Duplicate_Object_Key;
+			value.pos = p.curr_token.pos;
+			delete(key, p.allocator);
+			return;
+		}
+
+		obj[key] = elem;
+
+		if p.spec == Specification.JSON5 {
+			// Allow trailing commas
+			if allow_token(p, Kind.Comma) {
+				continue;
+			}
+		} else {
+			// Disallow trailing commas
+			if allow_token(p, Kind.Comma) {
+				continue;
+			} else {
+				break;
+			}
+		}
+	}
+
+	if err = expect_token(p, Kind.Close_Brace); err != Error.None {
+		value.pos = p.curr_token.pos;
+		return;
+	}
+
+	value.value = obj;
+	return;
+}
+
+
+// IMPORTANT NOTE(bill): unquote_string assumes a mostly valid string
+unquote_string :: proc(token: Token, spec: Specification, allocator := context.allocator) -> string {
+	get_u2_rune :: proc(s: string) -> rune {
+		if len(s) < 4 || s[0] != '\\' || s[1] != 'x' {
+			return -1;
+		}
+
+		r: rune;
+		for c in s[2:4] {
+			x: rune;
+			switch c {
+			case '0'..'9': x = c - '0';
+			case 'a'..'f': x = c - 'a' + 10;
+			case 'A'..'F': x = c - 'A' + 10;
+			case: return -1;
+			}
+			r = r*16 + x;
+		}
+		return r;
+	}
+	get_u4_rune :: proc(s: string) -> rune {
+		if len(s) < 6 || s[0] != '\\' || s[1] != 'u' {
+			return -1;
+		}
+
+		r: rune;
+		for c in s[2:6] {
+			x: rune;
+			switch c {
+			case '0'..'9': x = c - '0';
+			case 'a'..'f': x = c - 'a' + 10;
+			case 'A'..'F': x = c - 'A' + 10;
+			case: return -1;
+			}
+			r = r*16 + x;
+		}
+		return r;
+	}
+
+	if token.kind != Kind.String {
+		return "";
+	}
+	s := token.text;
+	if len(s) <= 2 {
+		return "";
+	}
+	quote := s[0];
+	if s[0] != s[len(s)-1] {
+		// Invalid string
+		return "";
+	}
+	s = s[1:len(s)-1];
+
+	i := 0;
+	for i < len(s) {
+		c := s[i];
+		if c == '\\' || c == quote || c < ' ' {
+			break;
+		}
+		if c < utf8.RUNE_SELF {
+			i += 1;
+			continue;
+		}
+		r, w := utf8.decode_rune_in_string(s);
+		if r == utf8.RUNE_ERROR && w == 1 {
+			break;
+		}
+		i += w;
+	}
+	if i == len(s) {
+		return clone_string(s, allocator);
+	}
+
+	b := make([]byte, len(s) + 2*utf8.UTF_MAX, allocator);
+	w := copy(b, cast([]byte)s[0:i]);
+	loop: for i < len(s) {
+		c := s[i];
+		switch {
+		case c == '\\':
+			i += 1;
+			if i >= len(s) {
+				break loop;
+			}
+			switch s[i] {
+			case: break loop;
+			case '"',  '\'', '\\', '/':
+				b[w] = s[i];
+				i += 1;
+				w += 1;
+
+			case 'b':
+				b[w] = '\b';
+				i += 1;
+				w += 1;
+			case 'f':
+				b[w] = '\f';
+				i += 1;
+				w += 1;
+			case 'r':
+				b[w] = '\r';
+				i += 1;
+				w += 1;
+			case 't':
+				b[w] = '\t';
+				i += 1;
+				w += 1;
+			case 'n':
+				b[w] = '\n';
+				i += 1;
+				w += 1;
+			case 'u':
+				i -= 1; // Include the \u in the check for sanity sake
+				r := get_u4_rune(s[i:]);
+				if r < 0 {
+					break loop;
+				}
+				i += 6;
+
+				buf, buf_width := utf8.encode_rune(r);
+				copy(b[w:], buf[:buf_width]);
+				w += buf_width;
+
+
+			case '0':
+				if spec == Specification.JSON5 {
+					b[w] = '\x00';
+					i += 1;
+					w += 1;
+				} else {
+					break loop;
+				}
+			case 'v':
+				if spec == Specification.JSON5 {
+					b[w] = '\v';
+					i += 1;
+					w += 1;
+				} else {
+					break loop;
+				}
+
+			case 'x':
+				if spec == Specification.JSON5 {
+					i -= 1; // Include the \x in the check for sanity sake
+					r := get_u2_rune(s[i:]);
+					if r < 0 {
+						break loop;
+					}
+					i += 4;
+
+					buf, buf_width := utf8.encode_rune(r);
+					copy(b[w:], buf[:buf_width]);
+					w += buf_width;
+				} else {
+					break loop;
+				}
+			}
+
+		case c == quote, c < ' ':
+			break loop;
+
+		case c < utf8.RUNE_SELF:
+			b[w] = c;
+			i += 1;
+			w += 1;
+
+		case:
+			r, width := utf8.decode_rune_in_string(s[i:]);
+			i += width;
+
+			buf, buf_width := utf8.encode_rune(r);
+			assert(buf_width <= width);
+			copy(b[w:], buf[:buf_width]);
+			w += buf_width;
+		}
+	}
+
+	return string(b[:w]);
+}

core/encoding/json/tokenizer.odin

@@ -0,0 +1,471 @@
+package json
+
+import "core:unicode/utf8"
+
+Token :: struct {
+	using pos: Pos,
+	kind: Kind,
+	text: string,
+}
+
+Kind :: enum {
+	Invalid,
+
+	Null,
+	False,
+	True,
+
+	Infinity,
+	NaN,
+
+	Ident,
+
+	Integer,
+	Float,
+	String,
+
+	Colon,
+	Comma,
+
+	Open_Brace,
+	Close_Brace,
+
+	Open_Bracket,
+	Close_Bracket,
+}
+
+Tokenizer :: struct {
+	using pos:        Pos,
+	data:             []byte,
+	r:                rune, // current rune
+	w:                int,  // current rune width in bytes
+	curr_line_offset: int,
+	spec:             Specification,
+}
+
+
+
+make_tokenizer :: proc(data: []byte, spec := Specification.JSON) -> Tokenizer {
+	t := Tokenizer{pos = {line=1}, data = data, spec = spec};
+	next_rune(&t);
+	if t.r == utf8.RUNE_BOM {
+		next_rune(&t);
+	}
+	return t;
+}
+
+next_rune :: proc(t: ^Tokenizer) -> rune #no_bounds_check {
+	if t.offset >= len(t.data) {
+		return utf8.RUNE_EOF;
+	}
+	t.offset += t.w;
+	t.r, t.w = utf8.decode_rune(t.data[t.offset:]);
+	t.pos.column = t.offset - t.curr_line_offset;
+	return t.r;
+}
+
+
+get_token :: proc(t: ^Tokenizer) -> (token: Token, err: Error) {
+	skip_digits :: proc(t: ^Tokenizer) {
+		for t.offset < len(t.data) {
+			if '0' <= t.r && t.r <= '9' {
+				// Okay
+			} else {
+				return;
+			}
+			next_rune(t);
+		}
+	}
+	skip_hex_digits :: proc(t: ^Tokenizer) {
+		for t.offset < len(t.data) {
+			next_rune(t);
+			switch t.r {
+			case '0'..'9', 'a'..'f', 'A'..'F':
+				// Okay
+			case:
+				return;
+			}
+		}
+	}
+
+	scan_espace :: proc(t: ^Tokenizer) -> bool {
+		switch t.r {
+		case '"', '\'', '\\', '/', 'b', 'n', 'r', 't', 'f':
+			next_rune(t);
+			return true;
+		case 'u':
+			// Expect 4 hexadecimal digits
+			for i := 0; i < 4; i += 1 {
+				r := next_rune(t);
+				switch r {
+				case '0'..'9', 'a'..'f', 'A'..'F':
+					// Okay
+				case:
+					return false;
+				}
+			}
+		case:
+			// Ignore the next rune regardless
+			next_rune(t);
+		}
+		return false;
+	}
+
+	skip_whitespace :: proc(t: ^Tokenizer) -> rune {
+		loop: for t.offset < len(t.data) {
+			switch t.r {
+			case ' ', '\t', '\v', '\f', '\r':
+				next_rune(t);
+			case '\n':
+				t.line += 1;
+				t.curr_line_offset = t.offset;
+				t.pos.column = 1;
+				next_rune(t);
+			case:
+				if t.spec == Specification.JSON5 {
+					switch t.r {
+					case 0x2028, 0x2029, 0xFEFF:
+						next_rune(t);
+						continue loop;
+					}
+				}
+				break loop;
+			}
+		}
+		return t.r;
+	}
+
+	skip_to_next_line :: proc(t: ^Tokenizer) {
+		for t.offset < len(t.data) {
+			r := next_rune(t);
+			if r == '\n' {
+				return;
+			}
+		}
+	}
+
+	skip_alphanum :: proc(t: ^Tokenizer) {
+		for t.offset < len(t.data) {
+			switch next_rune(t) {
+			case 'A'..'Z', 'a'..'z', '0'..'9', '_':
+				continue;
+			}
+
+			return;
+		}
+	}
+
+	skip_whitespace(t);
+
+	token.pos = t.pos;
+
+	token.kind = Kind.Invalid;
+
+	curr_rune := t.r;
+	next_rune(t);
+
+	block: switch curr_rune {
+	case utf8.RUNE_ERROR:
+		err = Error.Illegal_Character;
+	case utf8.RUNE_EOF, '\x00':
+		err = Error.EOF;
+
+	case 'A'..'Z', 'a'..'z', '_':
+		token.kind = Kind.Ident;
+
+		skip_alphanum(t);
+
+		switch str := string(t.data[token.offset:t.offset]); str {
+		case "null":  token.kind = Kind.Null;
+		case "false": token.kind = Kind.False;
+		case "true":  token.kind = Kind.True;
+		case:
+			if t.spec == Specification.JSON5 do switch str {
+			case "Infinity": token.kind = Kind.Infinity;
+			case "NaN":      token.kind = Kind.NaN;
+			}
+		}
+
+	case '+':
+		err = Error.Illegal_Character;
+		if t.spec != Specification.JSON5 {
+			break;
+		}
+		fallthrough;
+
+	case '-':
+		switch t.r {
+		case '0'..'9':
+			// Okay
+		case:
+			// Illegal use of +/-
+			err = Error.Illegal_Character;
+
+			if t.spec == Specification.JSON5 {
+				if t.r == 'I' || t.r == 'N' {
+					skip_alphanum(t);
+				}
+				switch string(t.data[token.offset:t.offset]) {
+				case "-Infinity": token.kind = Kind.Infinity;
+				case "-NaN":      token.kind = Kind.NaN;
+				}
+			}
+			break block;
+		}
+		fallthrough;
+
+	case '0'..'9':
+		token.kind = Kind.Integer;
+		if t.spec == Specification.JSON5 { // Hexadecimal Numbers
+			if curr_rune == '0' && (t.r == 'x' || t.r == 'X') {
+				next_rune(t);
+				skip_hex_digits(t);
+				break;
+			}
+		}
+
+		skip_digits(t);
+
+		if t.r == '.' {
+			token.kind = Kind.Float;
+			next_rune(t);
+			skip_digits(t);
+		}
+		if t.r == 'e' || t.r == 'E' {
+			switch r := next_rune(t); r {
+			case '+', '-':
+				next_rune(t);
+			}
+			skip_digits(t);
+		}
+
+		str := string(t.data[token.offset:t.offset]);
+		if !is_valid_number(str, t.spec) {
+			err = Error.Invalid_Number;
+		}
+
+	case '.':
+		err = Error.Illegal_Character;
+		if t.spec == Specification.JSON5 { // Allow leading decimal point
+			skip_digits(t);
+			if t.r == 'e' || t.r == 'E' {
+				switch r := next_rune(t); r {
+				case '+', '-':
+					next_rune(t);
+				}
+				skip_digits(t);
+			}
+			str := string(t.data[token.offset:t.offset]);
+			if !is_valid_number(str, t.spec) {
+				err = Error.Invalid_Number;
+			}
+		}
+
+
+	case '\'':
+		err = Error.Illegal_Character;
+		if t.spec != Specification.JSON5 {
+			break;
+		}
+		fallthrough;
+	case '"':
+		token.kind = Kind.String;
+		quote := curr_rune;
+		for t.offset < len(t.data) {
+			r := t.r;
+			if r == '\n' || r < 0 {
+				err = Error.String_Not_Terminated;
+				break;
+			}
+			next_rune(t);
+			if r == quote {
+				break;
+			}
+			if r == '\\' {
+				scan_espace(t);
+			}
+		}
+
+		str := string(t.data[token.offset : t.offset]);
+		if !is_valid_string_literal(str, t.spec) {
+			err = Error.Invalid_String;
+		}
+
+
+	case ',': token.kind = Kind.Comma;
+	case ':': token.kind = Kind.Colon;
+	case '{': token.kind = Kind.Open_Brace;
+	case '}': token.kind = Kind.Close_Brace;
+	case '[': token.kind = Kind.Open_Bracket;
+	case ']': token.kind = Kind.Close_Bracket;
+
+	case '/':
+		err = Error.Illegal_Character;
+		if t.spec == Specification.JSON5 {
+			switch t.r {
+			case '/':
+				// Single-line comments
+				skip_to_next_line(t);
+				return get_token(t);
+			case '*':
+				// None-nested multi-line comments
+				for t.offset < len(t.data) {
+					next_rune(t);
+					if t.r == '*' {
+						next_rune(t);
+						if t.r == '/' {
+							next_rune(t);
+							return get_token(t);
+						}
+					}
+				}
+				err = Error.EOF;
+			}
+		}
+
+	case: err = Error.Illegal_Character;
+	}
+
+	token.text = string(t.data[token.offset : t.offset]);
+
+	return;
+}
+
+
+
+is_valid_number :: proc(s: string, spec: Specification) -> bool {
+	if s == "" {
+		return false;
+	}
+
+	if s[0] == '-' {
+		s = s[1:];
+		if s == "" {
+			return false;
+		}
+	} else if spec == Specification.JSON5 {
+		if s[0] == '+' { // Allow positive sign
+			s = s[1:];
+			if s == "" {
+				return false;
+			}
+		}
+	}
+
+	switch s[0] {
+	case '0':
+		s = s[1:];
+	case '1'..'9':
+		s = s[1:];
+		for len(s) > 0 && '0' <= s[0] && s[0] <= '9' do s = s[1:];
+	case '.':
+		if spec == Specification.JSON5 { // Allow leading decimal point
+			s = s[1:];
+		} else {
+			return false;
+		}
+	case:
+		return false;
+	}
+
+	if spec == Specification.JSON5 {
+		if len(s) == 1 && s[0] == '.' { // Allow trailing decimal point
+			return true;
+		}
+	}
+
+	if len(s) >= 2 && s[0] == '.' && '0' <= s[1] && s[1] <= '9' {
+		s = s[2:];
+		for len(s) > 0 && '0' <= s[0] && s[0] <= '9' do s = s[1:];
+	}
+
+	if len(s) >= 2 && (s[0] == 'e' || s[0] == 'E') {
+		s = s[1:];
+		switch s[0] {
+		case '+', '-':
+			s = s[1:];
+			if s == "" {
+				return false;
+			}
+		}
+		for len(s) > 0 && '0' <= s[0] && s[0] <= '9' do s = s[1:];
+	}
+
+	// The string should be empty now to be valid
+	return s == "";
+}
+
+is_valid_string_literal :: proc(s: string, spec: Specification) -> bool {
+	if len(s) < 2 {
+		return false;
+	}
+	quote := s[0];
+	if s[0] != s[len(s)-1] {
+		return false;
+	}
+	if s[0] != '"' || s[len(s)-1] != '"' {
+		if spec == Specification.JSON5 {
+			if s[0] != '\'' || s[len(s)-1] != '\'' {
+				return false;
+			}
+		} else {
+			return false;
+		}
+	}
+	s = s[1 : len(s)-1];
+
+	i := 0;
+	for i < len(s) {
+		c := s[i];
+		switch {
+		case c == '\\':
+			i += 1;
+			if i >= len(s) {
+				return false;
+			}
+			switch s[i] {
+			case '"', '\'', '\\', '/', 'b', 'n', 'r', 't', 'f':
+				i += 1;
+			case 'u':
+				if i >= len(s) {
+					return false;
+				}
+				hex := s[i+1:];
+				if len(hex) < 4 {
+					return false;
+				}
+				hex = hex[:4];
+				i += 5;
+
+				for j := 0; j < 4; j += 1 {
+					c2 := hex[j];
+					switch c2 {
+					case '0'..'9', 'a'..'z', 'A'..'Z':
+						// Okay
+					case:
+						return false;
+					}
+				}
+
+			case: return false;
+			}
+
+		case c == quote, c < ' ':
+			return false;
+
+		case c < utf8.RUNE_SELF:
+			i += 1;
+
+		case:
+			r, width := utf8.decode_rune_in_string(s[i:]);
+			if r == utf8.RUNE_ERROR && width == 1 {
+				return false;
+			}
+			i += width;
+		}
+	}
+	if i == len(s) {
+		return true;
+	}
+	return true;
+}

core/encoding/json/types.odin

@@ -0,0 +1,73 @@
+package json
+
+Specification :: enum {
+	JSON,
+	JSON5,
+}
+
+Null    :: distinct rawptr;
+Integer :: i64;
+Float   :: f64;
+Boolean :: bool;
+String  :: string;
+Array   :: distinct [dynamic]Value;
+Object  :: distinct map[string]Value;
+
+Value :: struct {
+	pos, end: Pos,
+	value: union {
+		Null,
+		Integer,
+		Float,
+		Boolean,
+		String,
+		Array,
+		Object,
+	}
+}
+
+Pos :: struct {
+	offset: int,
+	line:   int,
+	column: int,
+}
+
+
+Error :: enum {
+	None,
+
+	EOF, // Not necessarily an error
+
+	// Tokenizing Errors
+	Illegal_Character,
+	Invalid_Number,
+	String_Not_Terminated,
+	Invalid_String,
+
+
+	// Parsing Errors
+	Unexpected_Token,
+	Expected_String_For_Object_Key,
+	Duplicate_Object_Key,
+	Expected_Colon_After_Key,
+}
+
+
+
+
+destroy_value :: proc(value: Value) {
+	switch v in value.value {
+	case Object:
+		for key, elem in v {
+			delete(key);
+			destroy_value(elem);
+		}
+		delete(v);
+	case Array:
+		for elem in v do destroy_value(elem);
+		delete(v);
+	case String:
+		delete(v);
+	}
+}
+

core/encoding/json/validator.odin

@@ -0,0 +1,123 @@
+package json
+
+import "core:mem"
+
+// NOTE(bill): is_valid will not check for duplicate keys
+is_valid :: proc(data: []byte, spec := Specification.JSON) -> bool {
+	p := make_parser(data, spec, mem.nil_allocator());
+	if p.spec == Specification.JSON5 {
+		return validate_value(&p);
+	}
+	return validate_object(&p);
+}
+
+validate_object_key :: proc(p: ^Parser) -> bool {
+	tok := p.curr_token;
+	if p.spec == Specification.JSON5 {
+		if tok.kind == Kind.String {
+			expect_token(p, Kind.String);
+			return true;
+		} else if tok.kind == Kind.Ident {
+			expect_token(p, Kind.Ident);
+			return true;
+		}
+	}
+	err := expect_token(p, Kind.String);
+	return err == Error.None;
+}
+validate_object :: proc(p: ^Parser) -> bool {
+	if err := expect_token(p, Kind.Open_Brace); err != Error.None {
+		return false;
+	}
+
+	for p.curr_token.kind != Kind.Close_Brace {
+		if !validate_object_key(p) {
+			return false;
+		}
+		if colon_err := expect_token(p, Kind.Colon); colon_err != Error.None {
+			return false;
+		}
+
+		if !validate_value(p) {
+			return false;
+		}
+
+		if p.spec == Specification.JSON5 {
+			// Allow trailing commas
+			if allow_token(p, Kind.Comma) {
+				continue;
+			}
+		} else {
+			// Disallow trailing commas
+			if allow_token(p, Kind.Comma) {
+				continue;
+			} else {
+				break;
+			}
+		}
+	}
+
+	if err := expect_token(p, Kind.Close_Brace); err != Error.None {
+		return false;
+	}
+	return true;
+}
+
+validate_array :: proc(p: ^Parser) -> bool {
+	if err := expect_token(p, Kind.Open_Bracket); err != Error.None {
+		return false;
+	}
+
+	for p.curr_token.kind != Kind.Close_Bracket {
+		if !validate_value(p) {
+			return false;
+		}
+
+		// Disallow trailing commas for the time being
+		if allow_token(p, Kind.Comma) {
+			continue;
+		} else {
+			break;
+		}
+	}
+
+	if err := expect_token(p, Kind.Close_Bracket); err != Error.None {
+		return false;
+	}
+
+	return true;
+}
+
+validate_value :: proc(p: ^Parser) -> bool {
+	token := p.curr_token;
+
+	using Kind;
+	switch token.kind {
+	case Null, False, True:
+		advance_token(p);
+		return true;
+	case Integer, Float:
+		advance_token(p);
+		return true;
+	case String:
+		advance_token(p);
+		return is_valid_string_literal(token.text, p.spec);
+
+	case Open_Brace:
+		return validate_object(p);
+
+	case Open_Bracket:
+		return validate_array(p);
+
+	case:
+		if p.spec == Specification.JSON5 {
+			switch token.kind {
+			case Infinity, NaN:
+				advance_token(p);
+				return true;
+			}
+		}
+	}
+
+	return false;
+}

core/hash/hash.odin

@@ -1,14 +1,28 @@
+package hash
+
+import "core:mem"
+
+adler32 :: proc(data: []byte) -> u32 {
+	ADLER_CONST :: 65521;
+	a, b: u32 = 1, 0;
+	for x in data {
+		a = (a + u32(x)) % ADLER_CONST;
+		b = (b + a) % ADLER_CONST;
+	}
+	return (b << 16) | a;
+}
+
 crc32 :: proc(data: []byte) -> u32 {
-	result := ~cast(u32)0;
+	result := ~u32(0);
 	for b in data {
-		result = result>>8 ~ __CRC32_TABLE[(result ~ cast(u32)b) & 0xff];
+		result = result>>8 ~ _crc32_table[(result ~ u32(b)) & 0xff];
 	}
 	return ~result;
 }
 crc64 :: proc(data: []byte) -> u64 {
-	result := ~cast(u64)0;
+	result := ~u64(0);
 	for b in data {
-		result = result>>8 ~ __CRC64_TABLE[(result ~ cast(u64)b) & 0xff];
+		result = result>>8 ~ _crc64_table[(result ~ u64(b)) & 0xff];
 	}
 	return ~result;
 }
@@ -16,7 +30,7 @@ crc64 :: proc(data: []byte) -> u64 {
 fnv32 :: proc(data: []byte) -> u32 {
 	h: u32 = 0x811c9dc5;
 	for b in data {
-		h = (h * 0x01000193) ~ cast(u32)b;
+		h = (h * 0x01000193) ~ u32(b);
 	}
 	return h;
 }
@@ -24,7 +38,7 @@ fnv32 :: proc(data: []byte) -> u32 {
 fnv64 :: proc(data: []byte) -> u64 {
 	h: u64 = 0xcbf29ce484222325;
 	for b in data {
-		h = (h * 0x100000001b3) ~ cast(u64)b;
+		h = (h * 0x100000001b3) ~ u64(b);
 	}
 	return h;
 }
@@ -32,7 +46,7 @@ fnv64 :: proc(data: []byte) -> u64 {
 fnv32a :: proc(data: []byte) -> u32 {
 	h: u32 = 0x811c9dc5;
 	for b in data {
-		h = (h ~ cast(u32)b) * 0x01000193;
+		h = (h ~ u32(b)) * 0x01000193;
 	}
 	return h;
 }
@@ -40,7 +54,7 @@ fnv32a :: proc(data: []byte) -> u32 {
 fnv64a :: proc(data: []byte) -> u64 {
 	h: u64 = 0xcbf29ce484222325;
 	for b in data {
-		h = (h ~ cast(u64)b) * 0x100000001b3;
+		h = (h ~ u64(b)) * 0x100000001b3;
 	}
 	return h;
 }
@@ -50,11 +64,11 @@ murmur32 :: proc(data: []byte) -> u32 {
 	c2_32: u32 : 0x1b873593;
 
 	h1: u32 = 0;
-	nblocks := data.count/4;
-	p := data.data;
-	p1 := p + 4*nblocks;
+	nblocks := len(data)/4;
+	p := &data[0];
+	p1 := mem.ptr_offset(p, 4*nblocks);
 
-	for ; p < p1; p += 4 {
+	for ; p < p1; p = mem.ptr_offset(p, 4) {
 		k1 := (cast(^u32)p)^;
 
 		k1 *= c1_32;
@@ -67,24 +81,23 @@ murmur32 :: proc(data: []byte) -> u32 {
 	}
 
 	tail := data[nblocks*4:];
-
 	k1: u32;
-	match tail.count&3 {
+	switch len(tail)&3 {
 	case 3:
-		k1 ~= cast(u32)tail[2] << 16;
+		k1 ~= u32(tail[2]) << 16;
 		fallthrough;
 	case 2:
-		k1 ~= cast(u32)tail[2] << 8;
+		k1 ~= u32(tail[2]) << 8;
 		fallthrough;
 	case 1:
-		k1 ~= cast(u32)tail[0];
+		k1 ~= u32(tail[0]);
 		k1 *= c1_32;
 		k1 = (k1 << 15) | (k1 >> 17) ;
 		k1 *= c2_32;
 		h1 ~= k1;
 	}
 
-	h1 ~= cast(u32)data.count;
+	h1 ~= u32(len(data));
 
 	h1 ~= h1 >> 16;
 	h1 *= 0x85ebca6b;
@@ -98,12 +111,12 @@ murmur32 :: proc(data: []byte) -> u32 {
 murmur64 :: proc(data: []byte) -> u64 {
 	SEED :: 0x9747b28c;
 
-	when false && size_of(int) == 8 {
+	when size_of(int) == 8 {
 		m :: 0xc6a4a7935bd1e995;
 		r :: 47;
 
-		h: u64 = SEED ~ (cast(u64)data.count * m);
-		data64 := slice_ptr(cast(^u64)^data[0], data.count/size_of(u64));
+		h: u64 = SEED ~ (u64(len(data)) * m);
+		data64 := mem.slice_ptr(cast(^u64)&data[0], len(data)/size_of(u64));
 
 		for _, i in data64 {
 			k := data64[i];
@@ -116,15 +129,15 @@ murmur64 :: proc(data: []byte) -> u64 {
 			h *= m;
 		}
 
-		match data.count&7 {
-		case 7: h ~= cast(u64)data[6] << 48; fallthrough;
-		case 6: h ~= cast(u64)data[5] << 40; fallthrough;
-		case 5: h ~= cast(u64)data[4] << 32; fallthrough;
-		case 4: h ~= cast(u64)data[3] << 24; fallthrough;
-		case 3: h ~= cast(u64)data[2] << 16; fallthrough;
-		case 2: h ~= cast(u64)data[1] << 8;  fallthrough;
+		switch len(data)&7 {
+		case 7: h ~= u64(data[6]) << 48; fallthrough;
+		case 6: h ~= u64(data[5]) << 40; fallthrough;
+		case 5: h ~= u64(data[4]) << 32; fallthrough;
+		case 4: h ~= u64(data[3]) << 24; fallthrough;
+		case 3: h ~= u64(data[2]) << 16; fallthrough;
+		case 2: h ~= u64(data[1]) << 8;  fallthrough;
 		case 1:
-			h ~= cast(u64)data[0];
+			h ~= u64(data[0]);
 			h *= m;
 		}
 
@@ -137,13 +150,12 @@ murmur64 :: proc(data: []byte) -> u64 {
 		m :: 0x5bd1e995;
 		r :: 24;
 
-		h1: u32 = cast(u32)SEED ~ cast(u32)data.count;
-		h2: u32 = SEED >> 32;
-
-		data32 := slice_ptr(cast(^u32)^data[0], data.count/size_of(u32));
-		len := data.count;
-
+		h1 := u32(SEED) ~ u32(len(data));
+		h2 := u32(SEED) >> 32;
+		data32 := mem.slice_ptr(cast(^u32)&data[0], len(data)/size_of(u32));
+		len := len(data);
 		i := 0;
+
 		for len >= 8 {
 			k1, k2: u32;
 			k1 = data32[i]; i += 1;
@@ -174,16 +186,17 @@ murmur64 :: proc(data: []byte) -> u64 {
 			len -= 4;
 		}
 
-		data8 := slice_to_bytes(data32[i:])[:3];
-		match len {
+		// TODO(bill): Fix this
+		#no_bounds_check data8 := mem.slice_to_bytes(data32[i:])[:3];
+		switch len {
 		case 3:
-			h2 ~= cast(u32)data8[2] << 16;
+			h2 ~= u32(data8[2]) << 16;
 			fallthrough;
 		case 2:
-			h2 ~= cast(u32)data8[1] << 8;
+			h2 ~= u32(data8[1]) << 8;
 			fallthrough;
 		case 1:
-			h2 ~= cast(u32)data8[0];
+			h2 ~= u32(data8[0]);
 			h2 *= m;
 		}
 
@@ -196,13 +209,12 @@ murmur64 :: proc(data: []byte) -> u64 {
 		h2 ~= h1>>19;
 		h2 *= m;
 
-		h := cast(u64)(h1)<<32 | cast(u64)(h2);
-		return h;
+		return u64(h1)<<32 | u64(h2);
 	}
 }
 
 
-__CRC32_TABLE := [256]u32{
+_crc32_table := [256]u32{
 	0x00000000, 0x77073096, 0xee0e612c, 0x990951ba,
 	0x076dc419, 0x706af48f, 0xe963a535, 0x9e6495a3,
 	0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988,
@@ -268,7 +280,7 @@ __CRC32_TABLE := [256]u32{
 	0xb3667a2e, 0xc4614ab8, 0x5d681b02, 0x2a6f2b94,
 	0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, 0x2d02ef8d,
 };
-__CRC64_TABLE := [256]u64{
+_crc64_table := [256]u64{
 	0x0000000000000000, 0x42f0e1eba9ea3693, 0x85e1c3d753d46d26, 0xc711223cfa3e5bb5,
 	0x493366450e42ecdf, 0x0bc387aea7a8da4c, 0xccd2a5925d9681f9, 0x8e224479f47cb76a,
 	0x9266cc8a1c85d9be, 0xd0962d61b56fef2d, 0x17870f5d4f51b498, 0x5577eeb6e6bb820b,

core/log/file_console_logger.odin

@@ -0,0 +1,143 @@
+package log
+
+import "core:fmt";
+import "core:strings";
+import "core:os";
+import "core:time";
+
+Level_Headers := []string{
+    "[DEBUG] --- ",
+    "[INFO ] --- ",
+    "[WARN ] --- ",
+    "[ERROR] --- ",
+    "[FATAL] --- ",
+};
+
+Default_Console_Logger_Opts :: Options{
+    Option.Level,
+    Option.Terminal_Color,
+    Option.Short_File_Path,
+    Option.Line,
+    Option.Procedure,
+} | Full_Timestamp_Opts;
+
+Default_File_Logger_Opts :: Options{
+    Option.Level,
+    Option.Short_File_Path,
+    Option.Line,
+    Option.Procedure,
+} | Full_Timestamp_Opts;
+
+
+File_Console_Logger_Data :: struct {
+    lowest_level: Level,
+    file_handle:  os.Handle,
+    ident : string,
+}
+
+create_file_logger :: proc(h: os.Handle, lowest := Level.Debug, opt := Default_File_Logger_Opts, ident := "") -> Logger {
+    data := new(File_Console_Logger_Data);
+    data.lowest_level = lowest;
+    data.file_handle = h;
+    data.ident = ident;
+    return Logger{file_console_logger_proc, data, opt};
+}
+
+destroy_file_logger ::proc(log : ^Logger) {
+    data := cast(^File_Console_Logger_Data)log.data;
+    if data.file_handle != os.INVALID_HANDLE do os.close(data.file_handle);
+    free(data);
+    log^ = nil_logger();
+}
+
+create_console_logger :: proc(lowest := Level.Debug, opt := Default_Console_Logger_Opts, ident := "") -> Logger {
+    data := new(File_Console_Logger_Data);
+    data.lowest_level = lowest;
+    data.file_handle = os.INVALID_HANDLE;
+    data.ident = ident;
+    return Logger{file_console_logger_proc, data, opt};
+}
+
+destroy_console_logger ::proc(log : ^Logger) {
+    free(log.data);
+    log^ = nil_logger();
+}
+
+file_console_logger_proc :: proc(logger_data: rawptr, level: Level, text: string, options: Options, location := #caller_location) {
+    data := cast(^File_Console_Logger_Data)logger_data;
+    if level < data.lowest_level do return;
+
+    h : os.Handle;
+    if(data.file_handle != os.INVALID_HANDLE) do h = data.file_handle;
+    else                                      do h = level <= Level.Error ? context.stdout : context.stderr;
+    backing: [1024]byte; //NOTE(Hoej): 1024 might be too much for a header backing, unless somebody has really long paths.
+    buf := strings.builder_from_slice(backing[:]);
+
+    do_level_header(options, level, &buf);
+
+    when time.IS_SUPPORTED {
+        if Full_Timestamp_Opts & options != nil {
+            fmt.sbprint(&buf, "[");
+            t := time.now();
+            y, m, d := time.date(t);
+            h, min, s := time.clock(t);
+            if Option.Date in options do fmt.sbprintf(&buf, "%d-%02d-%02d ", y, m, d);
+            if Option.Time in options do fmt.sbprintf(&buf, "%02d:%02d:%02d", h, min, s);
+            fmt.sbprint(&buf, "] ");
+        }
+    }
+
+    do_location_header(options, &buf, location);
+
+    if data.ident != "" do fmt.sbprintf(&buf, "[%s] ", data.ident);
+    //TODO(Hoej): When we have better atomics and such, make this thread-safe
+    fmt.fprintf(h, "%s %s\n", strings.to_string(buf), text);
+}
+
+do_level_header :: proc(opts : Options, level : Level, str : ^strings.Builder) {
+
+    RESET     :: "\x1b[0m";
+    RED       :: "\x1b[31m";
+    YELLOW    :: "\x1b[33m";
+    DARK_GREY :: "\x1b[90m";
+
+    col := RESET;
+    switch level {
+    case Level.Debug              : col = DARK_GREY;
+    case Level.Info               : col = RESET;
+    case Level.Warning            : col = YELLOW;
+    case Level.Error, Level.Fatal : col = RED;
+    }
+
+    if Option.Level in opts {
+        if Option.Terminal_Color in opts do fmt.sbprint(str, col);
+        fmt.sbprint(str, Level_Headers[level]);
+        if Option.Terminal_Color in opts do fmt.sbprint(str, RESET);
+    }
+}
+
+do_location_header :: proc(opts : Options, buf : ^strings.Builder, location := #caller_location) {
+    if Location_Header_Opts & opts != nil do fmt.sbprint(buf, "["); else do return;
+
+    file := location.file_path;
+    if Option.Short_File_Path in opts {
+        when os.OS == "windows" do delimiter := '\\'; else do delimiter := '/';
+        last := 0;
+        for r, i in location.file_path do if r == delimiter do last = i+1;
+        file = location.file_path[last:];
+    }
+
+    if Location_File_Opts & opts != nil do fmt.sbprint(buf, file);
+
+    if Option.Procedure in opts {
+        if Location_File_Opts & opts != nil do fmt.sbprint(buf, ".");
+        fmt.sbprintf(buf, "%s()", location.procedure);
+    }
+
+    if Option.Line in opts {
+        if Location_File_Opts & opts != nil || Option.Procedure in opts do fmt.sbprint(buf, ":");
+        fmt.sbprint(buf, location.line);
+    }
+
+    fmt.sbprint(buf, "] ");
+}

core/log/log.odin

@@ -0,0 +1,89 @@
+package log
+
+import "core:fmt"
+
+Level :: enum {
+	Debug,
+	Info,
+	Warning,
+	Error,
+	Fatal,
+}
+
+Option :: enum {
+    Level,
+    Date,
+    Time,
+    Short_File_Path,
+    Long_File_Path,
+    Line,
+    Procedure,
+    Terminal_Color
+}
+
+Options :: bit_set[Option];
+Full_Timestamp_Opts :: Options{
+    Option.Date,
+    Option.Time
+};
+Location_Header_Opts :: Options{
+    Option.Short_File_Path,
+    Option.Long_File_Path,
+    Option.Line,
+    Option.Procedure,
+};
+Location_File_Opts :: Options{
+    Option.Short_File_Path,
+    Option.Long_File_Path
+};
+
+Logger_Proc :: #type proc(data: rawptr, level: Level, text: string, options: Options, location := #caller_location);
+
+Logger :: struct {
+	procedure: Logger_Proc,
+	data:      rawptr,
+    options:   Options,
+}
+
+Multi_Logger_Data :: struct {
+    loggers : []Logger,
+}
+
+create_multi_logger :: proc(logs: ..Logger) -> Logger {
+    data := new(Multi_Logger_Data);
+    data.loggers = make([]Logger, len(logs));
+    copy(data.loggers, logs);
+    return Logger{multi_logger_proc, data, nil};
+}
+
+destroy_multi_logger ::proc(log : ^Logger) {
+    free(log.data);
+    log^ = nil_logger();
+}
+
+multi_logger_proc :: proc(logger_data: rawptr, level: Level, text: string,
+                          options: Options, location := #caller_location) {
+    data := cast(^Multi_Logger_Data)logger_data;
+    if data.loggers == nil || len(data.loggers) == 0 do return;
+    for log in data.loggers do log.procedure(log.data, level, text, log.options, location);
+}
+
+nil_logger_proc :: proc(data: rawptr, level: Level, text: string, options: Options, location := #caller_location) {
+	// Do nothing
+}
+
+nil_logger :: proc() -> Logger {
+	return Logger{nil_logger_proc, nil, nil};
+}
+
+debug :: proc(fmt_str : string, args : ..any, location := #caller_location) do logf(level=Level.Debug,   fmt_str=fmt_str, args=args, location=location);
+info  :: proc(fmt_str : string, args : ..any, location := #caller_location) do logf(level=Level.Info,    fmt_str=fmt_str, args=args, location=location);
+warn  :: proc(fmt_str : string, args : ..any, location := #caller_location) do logf(level=Level.Warning, fmt_str=fmt_str, args=args, location=location);
+error :: proc(fmt_str : string, args : ..any, location := #caller_location) do logf(level=Level.Error,   fmt_str=fmt_str, args=args, location=location);
+fatal :: proc(fmt_str : string, args : ..any, location := #caller_location) do logf(level=Level.Fatal,   fmt_str=fmt_str, args=args, location=location);
+
+logf :: proc(level : Level, fmt_str : string, args : ..any, location := #caller_location) {
+    logger := context.logger;
+    str := len(args) > 0 ? fmt.tprintf(fmt_str, ..args) : fmt.tprint(fmt_str); //NOTE(Hoej): While tprint isn't thread-safe, no logging is.
+    logger.procedure(logger.data, level, str, logger.options, location);
+}

core/math.odin

@@ -1,372 +0,0 @@
-TAU          :: 6.28318530717958647692528676655900576;
-PI           :: 3.14159265358979323846264338327950288;
-ONE_OVER_TAU :: 0.636619772367581343075535053490057448;
-ONE_OVER_PI  :: 0.159154943091895335768883763372514362;
-
-E            :: 2.71828182845904523536;
-SQRT_TWO     :: 1.41421356237309504880168872420969808;
-SQRT_THREE   :: 1.73205080756887729352744634150587236;
-SQRT_FIVE    :: 2.23606797749978969640917366873127623;
-
-LOG_TWO      :: 0.693147180559945309417232121458176568;
-LOG_TEN      :: 2.30258509299404568401799145468436421;
-
-EPSILON      :: 1.19209290e-7;
-
-τ :: TAU;
-π :: PI;
-
-Vec2 :: [vector 2]f32;
-Vec3 :: [vector 3]f32;
-Vec4 :: [vector 4]f32;
-
-Mat2 :: [2]Vec2;
-Mat3 :: [3]Vec3;
-Mat4 :: [4]Vec4;
-
-sqrt :: proc(x: f32) -> f32 #foreign __llvm_core "llvm.sqrt.f32";
-sqrt :: proc(x: f64) -> f64 #foreign __llvm_core "llvm.sqrt.f64";
-
-sin :: proc(x: f32) -> f32 #foreign __llvm_core "llvm.sin.f32";
-sin :: proc(x: f64) -> f64 #foreign __llvm_core "llvm.sin.f64";
-
-cos :: proc(x: f32) -> f32 #foreign __llvm_core "llvm.cos.f32";
-cos :: proc(x: f64) -> f64 #foreign __llvm_core "llvm.cos.f64";
-
-tan :: proc(x: f32) -> f32 #inline { return sin(x)/cos(x); }
-tan :: proc(x: f64) -> f64 #inline { return sin(x)/cos(x); }
-
-lerp :: proc(a, b, t: f32) -> f32 { return a*(1-t) + b*t; }
-lerp :: proc(a, b, t: f64) -> f64 { return a*(1-t) + b*t; }
-
-sign :: proc(x: f32) -> f32 { if x >= 0 { return +1; } return -1; }
-sign :: proc(x: f64) -> f64 { if x >= 0 { return +1; } return -1; }
-
-bit_reverse :: proc(b: u16) -> u16 #foreign __llvm_core "llvm.bitreverse.i16";
-bit_reverse :: proc(b: u32) -> u32 #foreign __llvm_core "llvm.bitreverse.i32";
-bit_reverse :: proc(b: u64) -> u64 #foreign __llvm_core "llvm.bitreverse.i64";
-
-fmuladd :: proc(a, b, c: f32) -> f32 #foreign __llvm_core "llvm.fmuladd.f32";
-fmuladd :: proc(a, b, c: f64) -> f64 #foreign __llvm_core "llvm.fmuladd.f64";
-
-
-copy_sign :: proc(x, y: f32) -> f32 {
-	ix := transmute(u32)x;
-	iy := transmute(u32)y;
-	ix &= 0x7fffffff;
-	ix |= iy & 0x80000000;
-	return transmute(f32)ix;
-}
-round :: proc(x: f32) -> f32 {
-	if x >= 0 {
-		return floor(x + 0.5);
-	}
-	return ceil(x - 0.5);
-}
-floor :: proc(x: f32) -> f32 {
-	if x >= 0 {
-		return cast(f32)cast(int)x;
-	}
-	return cast(f32)cast(int)(x-0.5);
-}
-ceil :: proc(x: f32) -> f32 {
-	if x < 0 {
-		return cast(f32)cast(int)x;
-	}
-	return cast(f32)cast(int)(x+1);
-}
-
-remainder32 :: proc(x, y: f32) -> f32 {
-	return x - round(x/y) * y;
-}
-
-fmod32 :: proc(x, y: f32) -> f32 {
-	y = abs(y);
-	result := remainder32(abs(x), y);
-	if sign(result) < 0 {
-		result += y;
-	}
-	return copy_sign(result, x);
-}
-
-
-to_radians :: proc(degrees: f32) -> f32 { return degrees * TAU / 360; }
-to_degrees :: proc(radians: f32) -> f32 { return radians * 360 / TAU; }
-
-
-
-
-dot :: proc(a, b: Vec2) -> f32 { c := a*b; return c.x + c.y; }
-dot :: proc(a, b: Vec3) -> f32 { c := a*b; return c.x + c.y + c.z; }
-dot :: proc(a, b: Vec4) -> f32 { c := a*b; return c.x + c.y + c.z + c.w; }
-
-cross :: proc(x, y: Vec3) -> Vec3 {
-	a := swizzle(x, 1, 2, 0) * swizzle(y, 2, 0, 1);
-	b := swizzle(x, 2, 0, 1) * swizzle(y, 1, 2, 0);
-	return a - b;
-}
-
-
-mag :: proc(v: Vec2) -> f32 { return sqrt(dot(v, v)); }
-mag :: proc(v: Vec3) -> f32 { return sqrt(dot(v, v)); }
-mag :: proc(v: Vec4) -> f32 { return sqrt(dot(v, v)); }
-
-norm :: proc(v: Vec2) -> Vec2 { return v / Vec2{mag(v)}; }
-norm :: proc(v: Vec3) -> Vec3 { return v / Vec3{mag(v)}; }
-norm :: proc(v: Vec4) -> Vec4 { return v / Vec4{mag(v)}; }
-
-norm0 :: proc(v: Vec2) -> Vec2 {
-	m := mag(v);
-	if m == 0 {
-		return Vec2{0};
-	}
-	return v / Vec2{m};
-}
-
-norm0 :: proc(v: Vec3) -> Vec3 {
-	m := mag(v);
-	if m == 0 {
-		return Vec3{0};
-	}
-	return v / Vec3{m};
-}
-
-norm0 :: proc(v: Vec4) -> Vec4 {
-	m := mag(v);
-	if m == 0 {
-		return Vec4{0};
-	}
-	return v / Vec4{m};
-}
-
-
-
-mat4_identity :: proc() -> Mat4 {
-	return Mat4{
-		{1, 0, 0, 0},
-		{0, 1, 0, 0},
-		{0, 0, 1, 0},
-		{0, 0, 0, 1},
-	};
-}
-
-mat4_transpose :: proc(m: Mat4) -> Mat4 {
-	for j in 0..<4 {
-		for i in 0..<4 {
-			m[i][j], m[j][i] = m[j][i], m[i][j];
-		}
-	}
-	return m;
-}
-
-mul :: proc(a, b: Mat4) -> Mat4 {
-	c: Mat4;
-	for j in 0..<4 {
-		for i in 0..<4 {
-			c[j][i] = a[0][i]*b[j][0] +
-			          a[1][i]*b[j][1] +
-			          a[2][i]*b[j][2] +
-			          a[3][i]*b[j][3];
-		}
-	}
-	return c;
-}
-
-mul :: proc(m: Mat4, v: Vec4) -> Vec4 {
-	return Vec4{
-		m[0][0]*v.x + m[1][0]*v.y + m[2][0]*v.z + m[3][0]*v.w,
-		m[0][1]*v.x + m[1][1]*v.y + m[2][1]*v.z + m[3][1]*v.w,
-		m[0][2]*v.x + m[1][2]*v.y + m[2][2]*v.z + m[3][2]*v.w,
-		m[0][3]*v.x + m[1][3]*v.y + m[2][3]*v.z + m[3][3]*v.w,
-	};
-}
-
-inverse :: proc(m: Mat4) -> Mat4 {
-	o: Mat4;
-
-	sf00 := m[2][2] * m[3][3] - m[3][2] * m[2][3];
-	sf01 := m[2][1] * m[3][3] - m[3][1] * m[2][3];
-	sf02 := m[2][1] * m[3][2] - m[3][1] * m[2][2];
-	sf03 := m[2][0] * m[3][3] - m[3][0] * m[2][3];
-	sf04 := m[2][0] * m[3][2] - m[3][0] * m[2][2];
-	sf05 := m[2][0] * m[3][1] - m[3][0] * m[2][1];
-	sf06 := m[1][2] * m[3][3] - m[3][2] * m[1][3];
-	sf07 := m[1][1] * m[3][3] - m[3][1] * m[1][3];
-	sf08 := m[1][1] * m[3][2] - m[3][1] * m[1][2];
-	sf09 := m[1][0] * m[3][3] - m[3][0] * m[1][3];
-	sf10 := m[1][0] * m[3][2] - m[3][0] * m[1][2];
-	sf11 := m[1][1] * m[3][3] - m[3][1] * m[1][3];
-	sf12 := m[1][0] * m[3][1] - m[3][0] * m[1][1];
-	sf13 := m[1][2] * m[2][3] - m[2][2] * m[1][3];
-	sf14 := m[1][1] * m[2][3] - m[2][1] * m[1][3];
-	sf15 := m[1][1] * m[2][2] - m[2][1] * m[1][2];
-	sf16 := m[1][0] * m[2][3] - m[2][0] * m[1][3];
-	sf17 := m[1][0] * m[2][2] - m[2][0] * m[1][2];
-	sf18 := m[1][0] * m[2][1] - m[2][0] * m[1][1];
-
-	o[0][0] = +(m[1][1] * sf00 - m[1][2] * sf01 + m[1][3] * sf02);
-	o[0][1] = -(m[1][0] * sf00 - m[1][2] * sf03 + m[1][3] * sf04);
-	o[0][2] = +(m[1][0] * sf01 - m[1][1] * sf03 + m[1][3] * sf05);
-	o[0][3] = -(m[1][0] * sf02 - m[1][1] * sf04 + m[1][2] * sf05);
-
-	o[1][0] = -(m[0][1] * sf00 - m[0][2] * sf01 + m[0][3] * sf02);
-	o[1][1] = +(m[0][0] * sf00 - m[0][2] * sf03 + m[0][3] * sf04);
-	o[1][2] = -(m[0][0] * sf01 - m[0][1] * sf03 + m[0][3] * sf05);
-	o[1][3] = +(m[0][0] * sf02 - m[0][1] * sf04 + m[0][2] * sf05);
-
-	o[2][0] = +(m[0][1] * sf06 - m[0][2] * sf07 + m[0][3] * sf08);
-	o[2][1] = -(m[0][0] * sf06 - m[0][2] * sf09 + m[0][3] * sf10);
-	o[2][2] = +(m[0][0] * sf11 - m[0][1] * sf09 + m[0][3] * sf12);
-	o[2][3] = -(m[0][0] * sf08 - m[0][1] * sf10 + m[0][2] * sf12);
-
-	o[3][0] = -(m[0][1] * sf13 - m[0][2] * sf14 + m[0][3] * sf15);
-	o[3][1] = +(m[0][0] * sf13 - m[0][2] * sf16 + m[0][3] * sf17);
-	o[3][2] = -(m[0][0] * sf14 - m[0][1] * sf16 + m[0][3] * sf18);
-	o[3][3] = +(m[0][0] * sf15 - m[0][1] * sf17 + m[0][2] * sf18);
-
-	ood := 1.0 / (m[0][0] * o[0][0] +
-	              m[0][1] * o[0][1] +
-	              m[0][2] * o[0][2] +
-	              m[0][3] * o[0][3]);
-
-	o[0][0] *= ood;
-	o[0][1] *= ood;
-	o[0][2] *= ood;
-	o[0][3] *= ood;
-	o[1][0] *= ood;
-	o[1][1] *= ood;
-	o[1][2] *= ood;
-	o[1][3] *= ood;
-	o[2][0] *= ood;
-	o[2][1] *= ood;
-	o[2][2] *= ood;
-	o[2][3] *= ood;
-	o[3][0] *= ood;
-	o[3][1] *= ood;
-	o[3][2] *= ood;
-	o[3][3] *= ood;
-
-	return o;
-}
-
-
-mat4_translate :: proc(v: Vec3) -> Mat4 {
-	m := mat4_identity();
-	m[3][0] = v.x;
-	m[3][1] = v.y;
-	m[3][2] = v.z;
-	m[3][3] = 1;
-	return m;
-}
-
-mat4_rotate :: proc(v: Vec3, angle_radians: f32) -> Mat4 {
-	c := cos(angle_radians);
-	s := sin(angle_radians);
-
-	a := norm(v);
-	t := a * Vec3{1-c};
-
-	rot := mat4_identity();
-
-	rot[0][0] = c + t.x*a.x;
-	rot[0][1] = 0 + t.x*a.y + s*a.z;
-	rot[0][2] = 0 + t.x*a.z - s*a.y;
-	rot[0][3] = 0;
-
-	rot[1][0] = 0 + t.y*a.x - s*a.z;
-	rot[1][1] = c + t.y*a.y;
-	rot[1][2] = 0 + t.y*a.z + s*a.x;
-	rot[1][3] = 0;
-
-	rot[2][0] = 0 + t.z*a.x + s*a.y;
-	rot[2][1] = 0 + t.z*a.y - s*a.x;
-	rot[2][2] = c + t.z*a.z;
-	rot[2][3] = 0;
-
-	return rot;
-}
-
-scale :: proc(m: Mat4, v: Vec3) -> Mat4 {
-	m[0][0] *= v.x;
-	m[1][1] *= v.y;
-	m[2][2] *= v.z;
-	return m;
-}
-
-scale :: proc(m: Mat4, s: f32) -> Mat4 {
-	m[0][0] *= s;
-	m[1][1] *= s;
-	m[2][2] *= s;
-	return m;
-}
-
-
-look_at :: proc(eye, centre, up: Vec3) -> Mat4 {
-	f := norm(centre - eye);
-	s := norm(cross(f, up));
-	u := cross(s, f);
-
-	m: Mat4;
-
-	m[0] = Vec4{+s.x, +s.y, +s.z, 0};
-	m[1] = Vec4{+u.x, +u.y, +u.z, 0};
-	m[2] = Vec4{-f.x, -f.y, -f.z, 0};
-	m[3] = Vec4{dot(s, eye), dot(u, eye), dot(f, eye), 1};
-
-	return m;
-}
-perspective :: proc(fovy, aspect, near, far: f32) -> Mat4 {
-	m: Mat4;
-	tan_half_fovy := tan(0.5 * fovy);
-	m[0][0] = 1.0 / (aspect*tan_half_fovy);
-	m[1][1] = 1.0 / (tan_half_fovy);
-	m[2][2] = -(far + near) / (far - near);
-	m[2][3] = -1.0;
-	m[3][2] = -2.0*far*near / (far - near);
-	return m;
-}
-
-
-ortho3d :: proc(left, right, bottom, top, near, far: f32) -> Mat4 {
-	m := mat4_identity();
-	m[0][0] = +2.0 / (right - left);
-	m[1][1] = +2.0 / (top - bottom);
-	m[2][2] = -2.0 / (far - near);
-	m[3][0] = -(right + left)   / (right - left);
-	m[3][1] = -(top   + bottom) / (top   - bottom);
-	m[3][2] = -(far + near) / (far - near);
-	return m;
-}
-
-
-
-
-
-F32_DIG        :: 6;
-F32_EPSILON    :: 1.192092896e-07;
-F32_GUARD      :: 0;
-F32_MANT_DIG   :: 24;
-F32_MAX        :: 3.402823466e+38;
-F32_MAX_10_EXP :: 38;
-F32_MAX_EXP    :: 128;
-F32_MIN        :: 1.175494351e-38;
-F32_MIN_10_EXP :: -37;
-F32_MIN_EXP    :: -125;
-F32_NORMALIZE  :: 0;
-F32_RADIX      :: 2;
-F32_ROUNDS     :: 1;
-
-F64_DIG        :: 15;                       // # of decimal digits of precision
-F64_EPSILON    :: 2.2204460492503131e-016;  // smallest such that 1.0+F64_EPSILON != 1.0
-F64_MANT_DIG   :: 53;                       // # of bits in mantissa
-F64_MAX        :: 1.7976931348623158e+308;  // max value
-F64_MAX_10_EXP :: 308;                      // max decimal exponent
-F64_MAX_EXP    :: 1024;                     // max binary exponent
-F64_MIN        :: 2.2250738585072014e-308;  // min positive value
-F64_MIN_10_EXP :: -307;                     // min decimal exponent
-F64_MIN_EXP    :: -1021;                    // min binary exponent
-F64_RADIX      :: 2;                        // exponent radix
-F64_ROUNDS     :: 1;                        // addition rounding: near
-
-
-

core/math/bits/bits.odin

@@ -0,0 +1,260 @@
+package bits
+
+import "core:os"
+
+U8_MIN  :: 0;
+U16_MIN :: 0;
+U32_MIN :: 0;
+U64_MIN :: 0;
+
+U8_MAX  :: 1 <<  8 - 1;
+U16_MAX :: 1 << 16 - 1;
+U32_MAX :: 1 << 32 - 1;
+U64_MAX :: 1 << 64 - 1;
+
+I8_MIN  :: - 1 << 7;
+I16_MIN :: - 1 << 15;
+I32_MIN :: - 1 << 31;
+I64_MIN :: - 1 << 63;
+
+I8_MAX  :: 1 <<  7 - 1;
+I16_MAX :: 1 << 15 - 1;
+I32_MAX :: 1 << 31 - 1;
+I64_MAX :: 1 << 63 - 1;
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.ctpop.i8")        count_ones8  :: proc(i:  u8) ->  u8 ---
+	@(link_name="llvm.ctpop.i16")       count_ones16 :: proc(i: u16) -> u16 ---
+	@(link_name="llvm.ctpop.i32")       count_ones32 :: proc(i: u32) -> u32 ---
+	@(link_name="llvm.ctpop.i64")       count_ones64 :: proc(i: u64) -> u64 ---
+
+	@(link_name="llvm.ctlz.i8")         leading_zeros8  :: proc(i:  u8, is_zero_undef := false) ->  u8 ---
+	@(link_name="llvm.ctlz.i16")        leading_zeros16 :: proc(i: u16, is_zero_undef := false) -> u16 ---
+	@(link_name="llvm.ctlz.i32")        leading_zeros32 :: proc(i: u32, is_zero_undef := false) -> u32 ---
+	@(link_name="llvm.ctlz.i64")        leading_zeros64 :: proc(i: u64, is_zero_undef := false) -> u64 ---
+
+	@(link_name="llvm.cttz.i8")         trailing_zeros8  :: proc(i:  u8,  is_zero_undef := false) ->  u8 ---
+	@(link_name="llvm.cttz.i16")        trailing_zeros16 :: proc(i: u16,  is_zero_undef := false) -> u16 ---
+	@(link_name="llvm.cttz.i32")        trailing_zeros32 :: proc(i: u32,  is_zero_undef := false) -> u32 ---
+	@(link_name="llvm.cttz.i64")        trailing_zeros64 :: proc(i: u64,  is_zero_undef := false) -> u64 ---
+
+	@(link_name="llvm.bitreverse.i8")   reverse_bits8  :: proc(i:  u8) ->  u8 ---
+	@(link_name="llvm.bitreverse.i16")  reverse_bits16 :: proc(i: u16) -> u16 ---
+	@(link_name="llvm.bitreverse.i32")  reverse_bits32 :: proc(i: u32) -> u32 ---
+	@(link_name="llvm.bitreverse.i64")  reverse_bits64 :: proc(i: u64) -> u64 ---
+
+	@(link_name="llvm.bswap.i16")       byte_swap_u16 :: proc(u16) -> u16 ---
+	@(link_name="llvm.bswap.i32")       byte_swap_u32 :: proc(u32) -> u32 ---
+	@(link_name="llvm.bswap.i64")       byte_swap_u64 :: proc(u64) -> u64 ---
+	@(link_name="llvm.bswap.i16")       byte_swap_i16 :: proc(i16) -> i16 ---
+	@(link_name="llvm.bswap.i32")       byte_swap_i32 :: proc(i32) -> i32 ---
+	@(link_name="llvm.bswap.i64")       byte_swap_i64 :: proc(i64) -> i64 ---
+	@(link_name="llvm.bswap.i128")      byte_swap_u128 :: proc(u128) -> u128 ---
+	@(link_name="llvm.bswap.i128")      byte_swap_i128 :: proc(i128) -> i128 ---
+}
+
+byte_swap_uint :: proc(i: uint) -> uint {
+	when size_of(uint) == size_of(u32) {
+		return uint(byte_swap_u32(u32(i)));
+	} else {
+		return uint(byte_swap_u64(u64(i)));
+	}
+}
+byte_swap_int :: proc(i: int) -> int {
+	when size_of(int) == size_of(i32) {
+		return int(byte_swap_i32(i32(i)));
+	} else {
+		return int(byte_swap_i64(i64(i)));
+	}
+}
+
+byte_swap :: proc{
+	byte_swap_u16,
+	byte_swap_u32,
+	byte_swap_u64,
+	byte_swap_u128,
+	byte_swap_i16,
+	byte_swap_i32,
+	byte_swap_i64,
+	byte_swap_i128,
+	byte_swap_uint,
+	byte_swap_int,
+};
+
+count_zeros8   :: proc(i:   u8) ->   u8 { return   8 - count_ones8(i); }
+count_zeros16  :: proc(i:  u16) ->  u16 { return  16 - count_ones16(i); }
+count_zeros32  :: proc(i:  u32) ->  u32 { return  32 - count_ones32(i); }
+count_zeros64  :: proc(i:  u64) ->  u64 { return  64 - count_ones64(i); }
+
+
+rotate_left8   :: proc(i: u8,   s: uint) ->   u8 { return (i << s)|(i >> (8*size_of(u8)   - s)); }
+rotate_left16  :: proc(i: u16,  s: uint) ->  u16 { return (i << s)|(i >> (8*size_of(u16)  - s)); }
+rotate_left32  :: proc(i: u32,  s: uint) ->  u32 { return (i << s)|(i >> (8*size_of(u32)  - s)); }
+rotate_left64  :: proc(i: u64,  s: uint) ->  u64 { return (i << s)|(i >> (8*size_of(u64)  - s)); }
+
+
+rotate_right8   :: proc(i: u8,   s: uint) ->   u8 { return (i >> s)|(i << (8*size_of(u8)   - s)); }
+rotate_right16  :: proc(i: u16,  s: uint) ->  u16 { return (i >> s)|(i << (8*size_of(u16)  - s)); }
+rotate_right32  :: proc(i: u32,  s: uint) ->  u32 { return (i >> s)|(i << (8*size_of(u32)  - s)); }
+rotate_right64  :: proc(i: u64,  s: uint) ->  u64 { return (i >> s)|(i << (8*size_of(u64)  - s)); }
+
+from_be_u8   :: proc(i:   u8) ->   u8 { return i; }
+from_be_u16  :: proc(i:  u16) ->  u16 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+from_be_u32  :: proc(i:  u32) ->  u32 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+from_be_u64  :: proc(i:  u64) ->  u64 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+from_be_uint :: proc(i: uint) -> uint { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+
+from_le_u8   :: proc(i:   u8) ->   u8 { return i; }
+from_le_u16  :: proc(i:  u16) ->  u16 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+from_le_u32  :: proc(i:  u32) ->  u32 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+from_le_u64  :: proc(i:  u64) ->  u64 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+from_le_uint :: proc(i: uint) -> uint { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+
+to_be_u8   :: proc(i:   u8) ->   u8 { return i; }
+to_be_u16  :: proc(i:  u16) ->  u16 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+to_be_u32  :: proc(i:  u32) ->  u32 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+to_be_u64  :: proc(i:  u64) ->  u64 { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+to_be_uint :: proc(i: uint) -> uint { when os.ENDIAN == "big" { return i; } else { return byte_swap(i); } }
+
+
+to_le_u8   :: proc(i:   u8) ->   u8 { return i; }
+to_le_u16  :: proc(i:  u16) ->  u16 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+to_le_u32  :: proc(i:  u32) ->  u32 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+to_le_u64  :: proc(i:  u64) ->  u64 { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+to_le_uint :: proc(i: uint) -> uint { when os.ENDIAN == "little" { return i; } else { return byte_swap(i); } }
+
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.uadd.with.overflow.i8")  overflowing_add_u8  :: proc(lhs, rhs:  u8) -> (u8, bool)  ---
+	@(link_name="llvm.sadd.with.overflow.i8")  overflowing_add_i8  :: proc(lhs, rhs:  i8) -> (i8, bool)  ---
+	@(link_name="llvm.uadd.with.overflow.i16") overflowing_add_u16 :: proc(lhs, rhs: u16) -> (u16, bool) ---
+	@(link_name="llvm.sadd.with.overflow.i16") overflowing_add_i16 :: proc(lhs, rhs: i16) -> (i16, bool) ---
+	@(link_name="llvm.uadd.with.overflow.i32") overflowing_add_u32 :: proc(lhs, rhs: u32) -> (u32, bool) ---
+	@(link_name="llvm.sadd.with.overflow.i32") overflowing_add_i32 :: proc(lhs, rhs: i32) -> (i32, bool) ---
+	@(link_name="llvm.uadd.with.overflow.i64") overflowing_add_u64 :: proc(lhs, rhs: u64) -> (u64, bool) ---
+	@(link_name="llvm.sadd.with.overflow.i64") overflowing_add_i64 :: proc(lhs, rhs: i64) -> (i64, bool) ---
+}
+
+overflowing_add_uint :: proc(lhs, rhs: uint) -> (uint, bool) {
+	when size_of(uint) == size_of(u32) {
+		x, ok := overflowing_add_u32(u32(lhs), u32(rhs));
+		return uint(x), ok;
+	} else {
+		x, ok := overflowing_add_u64(u64(lhs), u64(rhs));
+		return uint(x), ok;
+	}
+}
+overflowing_add_int :: proc(lhs, rhs: int) -> (int, bool) {
+	when size_of(int) == size_of(i32) {
+		x, ok := overflowing_add_i32(i32(lhs), i32(rhs));
+		return int(x), ok;
+	} else {
+		x, ok := overflowing_add_i64(i64(lhs), i64(rhs));
+		return int(x), ok;
+	}
+}
+
+overflowing_add :: proc{
+	overflowing_add_u8,   overflowing_add_i8,
+	overflowing_add_u16,  overflowing_add_i16,
+	overflowing_add_u32,  overflowing_add_i32,
+	overflowing_add_u64,  overflowing_add_i64,
+	overflowing_add_uint, overflowing_add_int,
+};
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.usub.with.overflow.i8")  overflowing_sub_u8  :: proc(lhs, rhs:  u8) -> (u8, bool)  ---
+	@(link_name="llvm.ssub.with.overflow.i8")  overflowing_sub_i8  :: proc(lhs, rhs:  i8) -> (i8, bool)  ---
+	@(link_name="llvm.usub.with.overflow.i16") overflowing_sub_u16 :: proc(lhs, rhs: u16) -> (u16, bool) ---
+	@(link_name="llvm.ssub.with.overflow.i16") overflowing_sub_i16 :: proc(lhs, rhs: i16) -> (i16, bool) ---
+	@(link_name="llvm.usub.with.overflow.i32") overflowing_sub_u32 :: proc(lhs, rhs: u32) -> (u32, bool) ---
+	@(link_name="llvm.ssub.with.overflow.i32") overflowing_sub_i32 :: proc(lhs, rhs: i32) -> (i32, bool) ---
+	@(link_name="llvm.usub.with.overflow.i64") overflowing_sub_u64 :: proc(lhs, rhs: u64) -> (u64, bool) ---
+	@(link_name="llvm.ssub.with.overflow.i64") overflowing_sub_i64 :: proc(lhs, rhs: i64) -> (i64, bool) ---
+}
+overflowing_sub_uint :: proc(lhs, rhs: uint) -> (uint, bool) {
+	when size_of(uint) == size_of(u32) {
+		x, ok := overflowing_sub_u32(u32(lhs), u32(rhs));
+		return uint(x), ok;
+	} else {
+		x, ok := overflowing_sub_u64(u64(lhs), u64(rhs));
+		return uint(x), ok;
+	}
+}
+overflowing_sub_int :: proc(lhs, rhs: int) -> (int, bool) {
+	when size_of(int) == size_of(i32) {
+		x, ok := overflowing_sub_i32(i32(lhs), i32(rhs));
+		return int(x), ok;
+	} else {
+		x, ok := overflowing_sub_i64(i64(lhs), i64(rhs));
+		return int(x), ok;
+	}
+}
+
+overflowing_sub :: proc{
+	overflowing_sub_u8,   overflowing_sub_i8,
+	overflowing_sub_u16,  overflowing_sub_i16,
+	overflowing_sub_u32,  overflowing_sub_i32,
+	overflowing_sub_u64,  overflowing_sub_i64,
+	overflowing_sub_uint, overflowing_sub_int,
+};
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.umul.with.overflow.i8")  overflowing_mul_u8  :: proc(lhs, rhs:  u8) -> (u8, bool)  ---
+	@(link_name="llvm.smul.with.overflow.i8")  overflowing_mul_i8  :: proc(lhs, rhs:  i8) -> (i8, bool)  ---
+	@(link_name="llvm.umul.with.overflow.i16") overflowing_mul_u16 :: proc(lhs, rhs: u16) -> (u16, bool) ---
+	@(link_name="llvm.smul.with.overflow.i16") overflowing_mul_i16 :: proc(lhs, rhs: i16) -> (i16, bool) ---
+	@(link_name="llvm.umul.with.overflow.i32") overflowing_mul_u32 :: proc(lhs, rhs: u32) -> (u32, bool) ---
+	@(link_name="llvm.smul.with.overflow.i32") overflowing_mul_i32 :: proc(lhs, rhs: i32) -> (i32, bool) ---
+	@(link_name="llvm.umul.with.overflow.i64") overflowing_mul_u64 :: proc(lhs, rhs: u64) -> (u64, bool) ---
+	@(link_name="llvm.smul.with.overflow.i64") overflowing_mul_i64 :: proc(lhs, rhs: i64) -> (i64, bool) ---
+}
+overflowing_mul_uint :: proc(lhs, rhs: uint) -> (uint, bool) {
+	when size_of(uint) == size_of(u32) {
+		x, ok := overflowing_mul_u32(u32(lhs), u32(rhs));
+		return uint(x), ok;
+	} else {
+		x, ok := overflowing_mul_u64(u64(lhs), u64(rhs));
+		return uint(x), ok;
+	}
+}
+overflowing_mul_int :: proc(lhs, rhs: int) -> (int, bool) {
+	when size_of(int) == size_of(i32) {
+		x, ok := overflowing_mul_i32(i32(lhs), i32(rhs));
+		return int(x), ok;
+	} else {
+		x, ok := overflowing_mul_i64(i64(lhs), i64(rhs));
+		return int(x), ok;
+	}
+}
+
+overflowing_mul :: proc{
+	overflowing_mul_u8,   overflowing_mul_i8,
+	overflowing_mul_u16,  overflowing_mul_i16,
+	overflowing_mul_u32,  overflowing_mul_i32,
+	overflowing_mul_u64,  overflowing_mul_i64,
+	overflowing_mul_uint, overflowing_mul_int,
+};
+
+is_power_of_two_u8   :: proc(i:   u8) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_i8   :: proc(i:   i8) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_u16  :: proc(i:  u16) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_i16  :: proc(i:  i16) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_u32  :: proc(i:  u32) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_i32  :: proc(i:  i32) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_u64  :: proc(i:  u64) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_i64  :: proc(i:  i64) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_uint :: proc(i: uint) -> bool { return i > 0 && (i & (i-1)) == 0; }
+is_power_of_two_int  :: proc(i:  int) -> bool { return i > 0 && (i & (i-1)) == 0; }
+
+is_power_of_two :: proc{
+	is_power_of_two_u8,   is_power_of_two_i8,
+	is_power_of_two_u16,  is_power_of_two_i16,
+	is_power_of_two_u32,  is_power_of_two_i32,
+	is_power_of_two_u64,  is_power_of_two_i64,
+	is_power_of_two_uint, is_power_of_two_int,
+};

core/math/math.odin

@@ -0,0 +1,598 @@
+package math
+
+TAU          :: 6.28318530717958647692528676655900576;
+PI           :: 3.14159265358979323846264338327950288;
+
+E            :: 2.71828182845904523536;
+SQRT_TWO     :: 1.41421356237309504880168872420969808;
+SQRT_THREE   :: 1.73205080756887729352744634150587236;
+SQRT_FIVE    :: 2.23606797749978969640917366873127623;
+
+LOG_TWO      :: 0.693147180559945309417232121458176568;
+LOG_TEN      :: 2.30258509299404568401799145468436421;
+
+EPSILON      :: 1.19209290e-7;
+
+τ :: TAU;
+π :: PI;
+
+Vec2 :: distinct [2]f32;
+Vec3 :: distinct [3]f32;
+Vec4 :: distinct [4]f32;
+
+// Column major
+Mat2 :: distinct [2][2]f32;
+Mat3 :: distinct [3][3]f32;
+Mat4 :: distinct [4][4]f32;
+
+Quat :: struct {x, y, z, w: f32};
+
+QUAT_IDENTITY := Quat{x = 0, y = 0, z = 0, w = 1};
+
+
+@(default_calling_convention="c")
+foreign _ {
+	@(link_name="llvm.sqrt.f32")
+	sqrt_f32 :: proc(x: f32) -> f32 ---;
+	@(link_name="llvm.sqrt.f64")
+	sqrt_f64 :: proc(x: f64) -> f64 ---;
+
+	@(link_name="llvm.sin.f32")
+	sin_f32 :: proc(θ: f32) -> f32 ---;
+	@(link_name="llvm.sin.f64")
+	sin_f64 :: proc(θ: f64) -> f64 ---;
+
+	@(link_name="llvm.cos.f32")
+	cos_f32 :: proc(θ: f32) -> f32 ---;
+	@(link_name="llvm.cos.f64")
+	cos_f64 :: proc(θ: f64) -> f64 ---;
+
+	@(link_name="llvm.pow.f32")
+	pow_f32 :: proc(x, power: f32) -> f32 ---;
+	@(link_name="llvm.pow.f64")
+	pow_f64 :: proc(x, power: f64) -> f64 ---;
+
+	@(link_name="llvm.fmuladd.f32")
+	fmuladd_f32 :: proc(a, b, c: f32) -> f32 ---;
+	@(link_name="llvm.fmuladd.f64")
+	fmuladd_f64 :: proc(a, b, c: f64) -> f64 ---;
+
+	@(link_name="llvm.log.f32")
+	log_f32 :: proc(x: f32) -> f32 ---;
+	@(link_name="llvm.log.f64")
+	log_f64 :: proc(x: f64) -> f64 ---;
+
+	@(link_name="llvm.exp.f32")
+	exp_f32 :: proc(x: f32) -> f32 ---;
+	@(link_name="llvm.exp.f64")
+	exp_f64 :: proc(x: f64) -> f64 ---;
+}
+
+log :: proc{log_f32, log_f64};
+exp :: proc{exp_f32, exp_f64};
+
+tan_f32 :: proc "c" (θ: f32) -> f32 { return sin(θ)/cos(θ); }
+tan_f64 :: proc "c" (θ: f64) -> f64 { return sin(θ)/cos(θ); }
+
+lerp :: proc(a, b: $T, t: $E) -> (x: T) { return a*(1-t) + b*t; }
+
+unlerp_f32 :: proc(a, b, x: f32) -> (t: f32) { return (x-a)/(b-a); }
+unlerp_f64 :: proc(a, b, x: f64) -> (t: f64) { return (x-a)/(b-a); }
+
+
+sign_f32 :: proc(x: f32) -> f32 { return x >= 0 ? +1 : -1; }
+sign_f64 :: proc(x: f64) -> f64 { return x >= 0 ? +1 : -1; }
+
+copy_sign_f32 :: proc(x, y: f32) -> f32 {
+	ix := transmute(u32)x;
+	iy := transmute(u32)y;
+	ix &= 0x7fff_ffff;
+	ix |= iy & 0x8000_0000;
+	return transmute(f32)ix;
+}
+
+copy_sign_f64 :: proc(x, y: f64) -> f64 {
+	ix := transmute(u64)x;
+	iy := transmute(u64)y;
+	ix &= 0x7fff_ffff_ffff_ffff;
+	ix |= iy & 0x8000_0000_0000_0000;
+	return transmute(f64)ix;
+}
+
+
+sqrt      :: proc{sqrt_f32, sqrt_f64};
+sin       :: proc{sin_f32, sin_f64};
+cos       :: proc{cos_f32, cos_f64};
+tan       :: proc{tan_f32, tan_f64};
+pow       :: proc{pow_f32, pow_f64};
+fmuladd   :: proc{fmuladd_f32, fmuladd_f64};
+sign      :: proc{sign_f32, sign_f64};
+copy_sign :: proc{copy_sign_f32, copy_sign_f64};
+
+
+round_f32 :: proc(x: f32) -> f32 { return x >= 0 ? floor(x + 0.5) : ceil(x - 0.5); }
+round_f64 :: proc(x: f64) -> f64 { return x >= 0 ? floor(x + 0.5) : ceil(x - 0.5); }
+round :: proc{round_f32, round_f64};
+
+floor_f32 :: proc(x: f32) -> f32 {
+	if x == 0 || is_nan(x) || is_inf(x) {
+		return x;
+	}
+	if x < 0 {
+		d, fract := modf(-x);
+		if fract != 0.0 {
+			d = d + 1;
+		}
+		return -d;
+	}
+	d, _ := modf(x);
+	return d;
+}
+floor_f64 :: proc(x: f64) -> f64 {
+	if x == 0 || is_nan(x) || is_inf(x) {
+		return x;
+	}
+	if x < 0 {
+		d, fract := modf(-x);
+		if fract != 0.0 {
+			d = d + 1;
+		}
+		return -d;
+	}
+	d, _ := modf(x);
+	return d;
+}
+floor :: proc{floor_f32, floor_f64};
+
+ceil_f32 :: proc(x: f32) -> f32 { return -floor_f32(-x); }
+ceil_f64 :: proc(x: f64) -> f64 { return -floor_f64(-x); }
+ceil :: proc{ceil_f32, ceil_f64};
+
+remainder_f32 :: proc(x, y: f32) -> f32 { return x - round(x/y) * y; }
+remainder_f64 :: proc(x, y: f64) -> f64 { return x - round(x/y) * y; }
+remainder :: proc{remainder_f32, remainder_f64};
+
+mod_f32 :: proc(x, y: f32) -> f32 {
+	result: f32;
+	y = abs(y);
+	result = remainder(abs(x), y);
+	if sign(result) < 0 {
+		result += y;
+	}
+	return copy_sign(result, x);
+}
+mod_f64 :: proc(x, y: f64) -> f64 {
+	result: f64;
+	y = abs(y);
+	result = remainder(abs(x), y);
+	if sign(result) < 0 {
+		result += y;
+	}
+	return copy_sign(result, x);
+}
+mod :: proc{mod_f32, mod_f64};
+
+// TODO(bill): These need to implemented with the actual instructions
+modf_f32 :: proc(x: f32) -> (int: f32, frac: f32) {
+	shift :: 32 - 8 - 1;
+	mask  :: 0xff;
+	bias  :: 127;
+
+	if x < 1 {
+		switch {
+		case x < 0:
+			int, frac = modf(-x);
+			return -int, -frac;
+		case x == 0:
+			return x, x;
+		}
+		return 0, x;
+	}
+
+	i := transmute(u32)x;
+	e := uint(i>>shift)&mask - bias;
+
+	if e < 32-9 {
+		i &~= 1<<(32-9-e) - 1;
+	}
+	int = transmute(f32)i;
+	frac = x - int;
+	return;
+}
+modf_f64 :: proc(x: f64) -> (int: f64, frac: f64) {
+	shift :: 64 - 11 - 1;
+	mask  :: 0x7ff;
+	bias  :: 1023;
+
+	if x < 1 {
+		switch {
+		case x < 0:
+			int, frac = modf(-x);
+			return -int, -frac;
+		case x == 0:
+			return x, x;
+		}
+		return 0, x;
+	}
+
+	i := transmute(u64)x;
+	e := uint(i>>shift)&mask - bias;
+
+	if e < 64-12 {
+		i &~= 1<<(64-12-e) - 1;
+	}
+	int = transmute(f64)i;
+	frac = x - int;
+	return;
+}
+modf :: proc{modf_f32, modf_f64};
+
+is_nan_f32 :: inline proc(x: f32) -> bool { return x != x; }
+is_nan_f64 :: inline proc(x: f64) -> bool { return x != x; }
+is_nan :: proc{is_nan_f32, is_nan_f64};
+
+is_finite_f32 :: inline proc(x: f32) -> bool { return !is_nan(x-x); }
+is_finite_f64 :: inline proc(x: f64) -> bool { return !is_nan(x-x); }
+is_finite :: proc{is_finite_f32, is_finite_f64};
+
+is_inf_f32 :: proc(x: f32, sign := 0) -> bool {
+	return sign >= 0 && x > F32_MAX || sign <= 0 && x < -F32_MAX;
+}
+is_inf_f64 :: proc(x: f64, sign := 0) -> bool {
+	return sign >= 0 && x > F64_MAX || sign <= 0 && x < -F64_MAX;
+}
+// If sign > 0,  is_inf reports whether f is positive infinity
+// If sign < 0,  is_inf reports whether f is negative infinity
+// If sign == 0, is_inf reports whether f is either   infinity
+is_inf :: proc{is_inf_f32, is_inf_f64};
+
+
+
+to_radians :: proc(degrees: f32) -> f32 { return degrees * TAU / 360; }
+to_degrees :: proc(radians: f32) -> f32 { return radians * 360 / TAU; }
+
+
+
+
+mul :: proc{
+	mat3_mul,
+	mat4_mul, mat4_mul_vec4,
+	quat_mul, quat_mulf,
+};
+
+div :: proc{quat_div, quat_divf};
+
+inverse :: proc{mat4_inverse, quat_inverse};
+dot     :: proc{vec_dot, quat_dot};
+cross   :: proc{cross2, cross3};
+
+vec_dot :: proc(a, b: $T/[$N]$E) -> E {
+	res: E;
+	for i in 0..<N {
+		res += a[i] * b[i];
+	}
+	return res;
+}
+
+cross2 :: proc(a, b: $T/[2]$E) -> E {
+	return a[0]*b[1] - a[1]*b[0];
+}
+
+cross3 :: proc(a, b: $T/[3]$E) -> T {
+	i := swizzle(a, 1, 2, 0) * swizzle(b, 2, 0, 1);
+	j := swizzle(a, 2, 0, 1) * swizzle(b, 1, 2, 0);
+	return T(i - j);
+}
+
+
+length :: proc(v: $T/[$N]$E) -> E { return sqrt(dot(v, v)); }
+
+norm :: proc(v: $T/[$N]$E) -> T { return v / length(v); }
+
+norm0 :: proc(v: $T/[$N]$E) -> T {
+	m := length(v);
+	return m == 0 ? 0 : v/m;
+}
+
+
+
+identity :: proc($T: typeid/[$N][N]$E) -> T {
+	m: T;
+	for i in 0..<N do m[i][i] = E(1);
+	return m;
+}
+
+transpose :: proc(m: $M/[$N][N]f32) -> M {
+	for j in 0..<N {
+		for i in 0..<N {
+			m[i][j], m[j][i] = m[j][i], m[i][j];
+		}
+	}
+	return m;
+}
+
+mat3_mul :: proc(a, b: Mat3) -> Mat3 {
+	c: Mat3;
+	for j in 0..<3 {
+		for i in 0..<3 {
+			c[j][i] = a[0][i]*b[j][0] +
+			          a[1][i]*b[j][1] +
+			          a[2][i]*b[j][2];
+		}
+	}
+	return c;
+}
+
+mat4_mul :: proc(a, b: Mat4) -> Mat4 {
+	c: Mat4;
+	for j in 0..<4 {
+		for i in 0..<4 {
+			c[j][i] = a[0][i]*b[j][0] +
+			          a[1][i]*b[j][1] +
+			          a[2][i]*b[j][2] +
+			          a[3][i]*b[j][3];
+		}
+	}
+	return c;
+}
+
+mat4_mul_vec4 :: proc(m: Mat4, v: Vec4) -> Vec4 {
+	return Vec4{
+		m[0][0]*v[0] + m[1][0]*v[1] + m[2][0]*v[2] + m[3][0]*v[3],
+		m[0][1]*v[0] + m[1][1]*v[1] + m[2][1]*v[2] + m[3][1]*v[3],
+		m[0][2]*v[0] + m[1][2]*v[1] + m[2][2]*v[2] + m[3][2]*v[3],
+		m[0][3]*v[0] + m[1][3]*v[1] + m[2][3]*v[2] + m[3][3]*v[3],
+	};
+}
+
+mat4_inverse :: proc(m: Mat4) -> Mat4 {
+	o: Mat4;
+
+	sf00 := m[2][2] * m[3][3] - m[3][2] * m[2][3];
+	sf01 := m[2][1] * m[3][3] - m[3][1] * m[2][3];
+	sf02 := m[2][1] * m[3][2] - m[3][1] * m[2][2];
+	sf03 := m[2][0] * m[3][3] - m[3][0] * m[2][3];
+	sf04 := m[2][0] * m[3][2] - m[3][0] * m[2][2];
+	sf05 := m[2][0] * m[3][1] - m[3][0] * m[2][1];
+	sf06 := m[1][2] * m[3][3] - m[3][2] * m[1][3];
+	sf07 := m[1][1] * m[3][3] - m[3][1] * m[1][3];
+	sf08 := m[1][1] * m[3][2] - m[3][1] * m[1][2];
+	sf09 := m[1][0] * m[3][3] - m[3][0] * m[1][3];
+	sf10 := m[1][0] * m[3][2] - m[3][0] * m[1][2];
+	sf11 := m[1][1] * m[3][3] - m[3][1] * m[1][3];
+	sf12 := m[1][0] * m[3][1] - m[3][0] * m[1][1];
+	sf13 := m[1][2] * m[2][3] - m[2][2] * m[1][3];
+	sf14 := m[1][1] * m[2][3] - m[2][1] * m[1][3];
+	sf15 := m[1][1] * m[2][2] - m[2][1] * m[1][2];
+	sf16 := m[1][0] * m[2][3] - m[2][0] * m[1][3];
+	sf17 := m[1][0] * m[2][2] - m[2][0] * m[1][2];
+	sf18 := m[1][0] * m[2][1] - m[2][0] * m[1][1];
+
+
+	o[0][0] = +(m[1][1] * sf00 - m[1][2] * sf01 + m[1][3] * sf02);
+	o[0][1] = -(m[1][0] * sf00 - m[1][2] * sf03 + m[1][3] * sf04);
+	o[0][2] = +(m[1][0] * sf01 - m[1][1] * sf03 + m[1][3] * sf05);
+	o[0][3] = -(m[1][0] * sf02 - m[1][1] * sf04 + m[1][2] * sf05);
+
+	o[1][0] = -(m[0][1] * sf00 - m[0][2] * sf01 + m[0][3] * sf02);
+	o[1][1] = +(m[0][0] * sf00 - m[0][2] * sf03 + m[0][3] * sf04);
+	o[1][2] = -(m[0][0] * sf01 - m[0][1] * sf03 + m[0][3] * sf05);
+	o[1][3] = +(m[0][0] * sf02 - m[0][1] * sf04 + m[0][2] * sf05);
+
+	o[2][0] = +(m[0][1] * sf06 - m[0][2] * sf07 + m[0][3] * sf08);
+	o[2][1] = -(m[0][0] * sf06 - m[0][2] * sf09 + m[0][3] * sf10);
+	o[2][2] = +(m[0][0] * sf11 - m[0][1] * sf09 + m[0][3] * sf12);
+	o[2][3] = -(m[0][0] * sf08 - m[0][1] * sf10 + m[0][2] * sf12);
+
+	o[3][0] = -(m[0][1] * sf13 - m[0][2] * sf14 + m[0][3] * sf15);
+	o[3][1] = +(m[0][0] * sf13 - m[0][2] * sf16 + m[0][3] * sf17);
+	o[3][2] = -(m[0][0] * sf14 - m[0][1] * sf16 + m[0][3] * sf18);
+	o[3][3] = +(m[0][0] * sf15 - m[0][1] * sf17 + m[0][2] * sf18);
+
+	ood := 1.0 / (m[0][0] * o[0][0] +
+	              m[0][1] * o[0][1] +
+	              m[0][2] * o[0][2] +
+	              m[0][3] * o[0][3]);
+
+	o[0][0] *= ood;
+	o[0][1] *= ood;
+	o[0][2] *= ood;
+	o[0][3] *= ood;
+	o[1][0] *= ood;
+	o[1][1] *= ood;
+	o[1][2] *= ood;
+	o[1][3] *= ood;
+	o[2][0] *= ood;
+	o[2][1] *= ood;
+	o[2][2] *= ood;
+	o[2][3] *= ood;
+	o[3][0] *= ood;
+	o[3][1] *= ood;
+	o[3][2] *= ood;
+	o[3][3] *= ood;
+
+	return o;
+}
+
+
+mat4_translate :: proc(v: Vec3) -> Mat4 {
+	m := identity(Mat4);
+	m[3][0] = v[0];
+	m[3][1] = v[1];
+	m[3][2] = v[2];
+	m[3][3] = 1;
+	return m;
+}
+
+mat4_rotate :: proc(v: Vec3, angle_radians: f32) -> Mat4 {
+	c := cos(angle_radians);
+	s := sin(angle_radians);
+
+	a := norm(v);
+	t := a * (1-c);
+
+	rot := identity(Mat4);
+
+	rot[0][0] = c + t[0]*a[0];
+	rot[0][1] = 0 + t[0]*a[1] + s*a[2];
+	rot[0][2] = 0 + t[0]*a[2] - s*a[1];
+	rot[0][3] = 0;
+
+	rot[1][0] = 0 + t[1]*a[0] - s*a[2];
+	rot[1][1] = c + t[1]*a[1];
+	rot[1][2] = 0 + t[1]*a[2] + s*a[0];
+	rot[1][3] = 0;
+
+	rot[2][0] = 0 + t[2]*a[0] + s*a[1];
+	rot[2][1] = 0 + t[2]*a[1] - s*a[0];
+	rot[2][2] = c + t[2]*a[2];
+	rot[2][3] = 0;
+
+	return rot;
+}
+
+scale_vec3 :: proc(m: Mat4, v: Vec3) -> Mat4 {
+	m[0][0] *= v[0];
+	m[1][1] *= v[1];
+	m[2][2] *= v[2];
+	return m;
+}
+
+scale_f32 :: proc(m: Mat4, s: f32) -> Mat4 {
+	m[0][0] *= s;
+	m[1][1] *= s;
+	m[2][2] *= s;
+	return m;
+}
+
+scale :: proc{scale_vec3, scale_f32};
+
+
+look_at :: proc(eye, centre, up: Vec3) -> Mat4 {
+	f := norm(centre - eye);
+	s := norm(cross(f, up));
+	u := cross(s, f);
+
+	return Mat4{
+		{+s.x, +u.x, -f.x, 0},
+		{+s.y, +u.y, -f.y, 0},
+		{+s.z, +u.z, -f.z, 0},
+		{-dot(s, eye), -dot(u, eye), dot(f, eye), 1},
+	};
+}
+
+perspective :: proc(fovy, aspect, near, far: f32) -> Mat4 {
+	m: Mat4;
+	tan_half_fovy := tan(0.5 * fovy);
+
+	m[0][0] = 1.0 / (aspect*tan_half_fovy);
+	m[1][1] = 1.0 / (tan_half_fovy);
+	m[2][2] = -(far + near) / (far - near);
+	m[2][3] = -1.0;
+	m[3][2] = -2.0*far*near / (far - near);
+	return m;
+}
+
+
+ortho3d :: proc(left, right, bottom, top, near, far: f32) -> Mat4 {
+	m := identity(Mat4);
+	m[0][0] = +2.0 / (right - left);
+	m[1][1] = +2.0 / (top - bottom);
+	m[2][2] = -2.0 / (far - near);
+	m[3][0] = -(right + left)   / (right - left);
+	m[3][1] = -(top   + bottom) / (top   - bottom);
+	m[3][2] = -(far + near) / (far - near);
+	return m;
+}
+
+
+// Quaternion operations
+
+conj :: proc(q: Quat) -> Quat {
+	return Quat{-q.x, -q.y, -q.z, q.w};
+}
+
+quat_mul :: proc(q0, q1: Quat) -> Quat {
+	d: Quat;
+	d.x = q0.w * q1.x + q0.x * q1.w + q0.y * q1.z - q0.z * q1.y;
+	d.y = q0.w * q1.y - q0.x * q1.z + q0.y * q1.w + q0.z * q1.x;
+	d.z = q0.w * q1.z + q0.x * q1.y - q0.y * q1.x + q0.z * q1.w;
+	d.w = q0.w * q1.w - q0.x * q1.x - q0.y * q1.y - q0.z * q1.z;
+	return d;
+}
+
+quat_mulf :: proc(q: Quat, f: f32) -> Quat { return Quat{q.x*f, q.y*f, q.z*f, q.w*f}; }
+quat_divf :: proc(q: Quat, f: f32) -> Quat { return Quat{q.x/f, q.y/f, q.z/f, q.w/f}; }
+
+quat_div     :: proc(q0, q1: Quat) -> Quat { return mul(q0, quat_inverse(q1)); }
+quat_inverse :: proc(q: Quat) -> Quat { return div(conj(q), dot(q, q)); }
+quat_dot     :: proc(q0, q1: Quat) -> f32 { return q0.x*q1.x + q0.y*q1.y + q0.z*q1.z + q0.w*q1.w; }
+
+quat_norm :: proc(q: Quat) -> Quat {
+	m := sqrt(dot(q, q));
+	return div(q, m);
+}
+
+axis_angle :: proc(axis: Vec3, angle_radians: f32) -> Quat {
+	v := norm(axis) * sin(0.5*angle_radians);
+	w := cos(0.5*angle_radians);
+	return Quat{v.x, v.y, v.z, w};
+}
+
+euler_angles :: proc(pitch, yaw, roll: f32) -> Quat {
+	p := axis_angle(Vec3{1, 0, 0}, pitch);
+	y := axis_angle(Vec3{0, 1, 0}, yaw);
+	r := axis_angle(Vec3{0, 0, 1}, roll);
+	return mul(mul(y, p), r);
+}
+
+quat_to_mat4 :: proc(q: Quat) -> Mat4 {
+	a := quat_norm(q);
+	xx := a.x*a.x; yy := a.y*a.y; zz := a.z*a.z;
+	xy := a.x*a.y; xz := a.x*a.z; yz := a.y*a.z;
+	wx := a.w*a.x; wy := a.w*a.y; wz := a.w*a.z;
+
+	m := identity(Mat4);
+
+	m[0][0] = 1 - 2*(yy + zz);
+	m[0][1] =     2*(xy + wz);
+	m[0][2] =     2*(xz - wy);
+
+	m[1][0] =     2*(xy - wz);
+	m[1][1] = 1 - 2*(xx + zz);
+	m[1][2] =     2*(yz + wx);
+
+	m[2][0] =     2*(xz + wy);
+	m[2][1] =     2*(yz - wx);
+	m[2][2] = 1 - 2*(xx + yy);
+	return m;
+}
+
+
+
+
+F32_DIG        :: 6;
+F32_EPSILON    :: 1.192092896e-07;
+F32_GUARD      :: 0;
+F32_MANT_DIG   :: 24;
+F32_MAX        :: 3.402823466e+38;
+F32_MAX_10_EXP :: 38;
+F32_MAX_EXP    :: 128;
+F32_MIN        :: 1.175494351e-38;
+F32_MIN_10_EXP :: -37;
+F32_MIN_EXP    :: -125;
+F32_NORMALIZE  :: 0;
+F32_RADIX      :: 2;
+F32_ROUNDS     :: 1;
+
+F64_DIG        :: 15;                       // # of decimal digits of precision
+F64_EPSILON    :: 2.2204460492503131e-016;  // smallest such that 1.0+F64_EPSILON != 1.0
+F64_MANT_DIG   :: 53;                       // # of bits in mantissa
+F64_MAX        :: 1.7976931348623158e+308;  // max value
+F64_MAX_10_EXP :: 308;                      // max decimal exponent
+F64_MAX_EXP    :: 1024;                     // max binary exponent
+F64_MIN        :: 2.2250738585072014e-308;  // min positive value
+F64_MIN_10_EXP :: -307;                     // min decimal exponent
+F64_MIN_EXP    :: -1021;                    // min binary exponent
+F64_RADIX      :: 2;                        // exponent radix
+F64_ROUNDS     :: 1;                        // addition rounding: near

core/math/rand/normal.odin

@@ -0,0 +1,146 @@
+package rand
+
+import "core:math"
+
+//
+//    Normal distribution
+//
+//    "The Ziggurat Method for Generating Random Variables"
+//    Authors: George Marsaglia, Wai Wan Tsang
+//    Submitted: 2000-04-15. Published: 2000-10-02.
+//    https://www.jstatsoft.org/index.php/jss/article/view/v005i08/ziggurat.pdf [pdf]
+//    https://www.jstatsoft.org/article/view/v005i08 [web page]
+//
+
+// norm_float64 returns a normally distributed f64 in the range -max(f64) through +max(f64) inclusive,
+// with a standard normal distribution with a mean of 0 and standard deviation of 1.
+//
+// sample = norm_float64() * std_dev + mean
+//
+norm_float64 :: proc(r: ^Rand = global_rand_ptr) -> f64 {
+	rn :: 3.442619855899;
+
+	@(static)
+	kn := [128]u32{
+		0x76ad2212, 0x00000000, 0x600f1b53, 0x6ce447a6, 0x725b46a2,
+		0x7560051d, 0x774921eb, 0x789a25bd, 0x799045c3, 0x7a4bce5d,
+		0x7adf629f, 0x7b5682a6, 0x7bb8a8c6, 0x7c0ae722, 0x7c50cce7,
+		0x7c8cec5b, 0x7cc12cd6, 0x7ceefed2, 0x7d177e0b, 0x7d3b8883,
+		0x7d5bce6c, 0x7d78dd64, 0x7d932886, 0x7dab0e57, 0x7dc0dd30,
+		0x7dd4d688, 0x7de73185, 0x7df81cea, 0x7e07c0a3, 0x7e163efa,
+		0x7e23b587, 0x7e303dfd, 0x7e3beec2, 0x7e46db77, 0x7e51155d,
+		0x7e5aabb3, 0x7e63abf7, 0x7e6c222c, 0x7e741906, 0x7e7b9a18,
+		0x7e82adfa, 0x7e895c63, 0x7e8fac4b, 0x7e95a3fb, 0x7e9b4924,
+		0x7ea0a0ef, 0x7ea5b00d, 0x7eaa7ac3, 0x7eaf04f3, 0x7eb3522a,
+		0x7eb765a5, 0x7ebb4259, 0x7ebeeafd, 0x7ec2620a, 0x7ec5a9c4,
+		0x7ec8c441, 0x7ecbb365, 0x7ece78ed, 0x7ed11671, 0x7ed38d62,
+		0x7ed5df12, 0x7ed80cb4, 0x7eda175c, 0x7edc0005, 0x7eddc78e,
+		0x7edf6ebf, 0x7ee0f647, 0x7ee25ebe, 0x7ee3a8a9, 0x7ee4d473,
+		0x7ee5e276, 0x7ee6d2f5, 0x7ee7a620, 0x7ee85c10, 0x7ee8f4cd,
+		0x7ee97047, 0x7ee9ce59, 0x7eea0eca, 0x7eea3147, 0x7eea3568,
+		0x7eea1aab, 0x7ee9e071, 0x7ee98602, 0x7ee90a88, 0x7ee86d08,
+		0x7ee7ac6a, 0x7ee6c769, 0x7ee5bc9c, 0x7ee48a67, 0x7ee32efc,
+		0x7ee1a857, 0x7edff42f, 0x7ede0ffa, 0x7edbf8d9, 0x7ed9ab94,
+		0x7ed7248d, 0x7ed45fae, 0x7ed1585c, 0x7ece095f, 0x7eca6ccb,
+		0x7ec67be2, 0x7ec22eee, 0x7ebd7d1a, 0x7eb85c35, 0x7eb2c075,
+		0x7eac9c20, 0x7ea5df27, 0x7e9e769f, 0x7e964c16, 0x7e8d44ba,
+		0x7e834033, 0x7e781728, 0x7e6b9933, 0x7e5d8a1a, 0x7e4d9ded,
+		0x7e3b737a, 0x7e268c2f, 0x7e0e3ff5, 0x7df1aa5d, 0x7dcf8c72,
+		0x7da61a1e, 0x7d72a0fb, 0x7d30e097, 0x7cd9b4ab, 0x7c600f1a,
+		0x7ba90bdc, 0x7a722176, 0x77d664e5,
+	};
+
+	@(static)
+	wn := [128]f32{
+		1.7290405e-09, 1.2680929e-10, 1.6897518e-10, 1.9862688e-10,
+		2.2232431e-10, 2.4244937e-10, 2.601613e-10,  2.7611988e-10,
+		2.9073963e-10, 3.042997e-10,  3.1699796e-10, 3.289802e-10,
+		3.4035738e-10, 3.5121603e-10, 3.616251e-10,  3.7164058e-10,
+		3.8130857e-10, 3.9066758e-10, 3.9975012e-10, 4.08584e-10,
+		4.1719309e-10, 4.2559822e-10, 4.338176e-10,  4.418672e-10,
+		4.497613e-10,  4.5751258e-10, 4.651324e-10,  4.7263105e-10,
+		4.8001775e-10, 4.87301e-10,   4.944885e-10,  5.015873e-10,
+		5.0860405e-10, 5.155446e-10,  5.2241467e-10, 5.2921934e-10,
+		5.359635e-10,  5.426517e-10,  5.4928817e-10, 5.5587696e-10,
+		5.624219e-10,  5.6892646e-10, 5.753941e-10,  5.818282e-10,
+		5.882317e-10,  5.946077e-10,  6.00959e-10,   6.072884e-10,
+		6.135985e-10,  6.19892e-10,   6.2617134e-10, 6.3243905e-10,
+		6.386974e-10,  6.449488e-10,  6.511956e-10,  6.5744005e-10,
+		6.6368433e-10, 6.699307e-10,  6.7618144e-10, 6.824387e-10,
+		6.8870465e-10, 6.949815e-10,  7.012715e-10,  7.075768e-10,
+		7.1389966e-10, 7.202424e-10,  7.266073e-10,  7.329966e-10,
+		7.394128e-10,  7.4585826e-10, 7.5233547e-10, 7.58847e-10,
+		7.653954e-10,  7.719835e-10,  7.7861395e-10, 7.852897e-10,
+		7.920138e-10,  7.987892e-10,  8.0561924e-10, 8.125073e-10,
+		8.194569e-10,  8.2647167e-10, 8.3355556e-10, 8.407127e-10,
+		8.479473e-10,  8.55264e-10,   8.6266755e-10, 8.7016316e-10,
+		8.777562e-10,  8.8545243e-10, 8.932582e-10,  9.0117996e-10,
+		9.09225e-10,   9.174008e-10,  9.2571584e-10, 9.341788e-10,
+		9.427997e-10,  9.515889e-10,  9.605579e-10,  9.697193e-10,
+		9.790869e-10,  9.88676e-10,   9.985036e-10,  1.0085882e-09,
+		1.0189509e-09, 1.0296151e-09, 1.0406069e-09, 1.0519566e-09,
+		1.063698e-09,  1.0758702e-09, 1.0885183e-09, 1.1016947e-09,
+		1.1154611e-09, 1.1298902e-09, 1.1450696e-09, 1.1611052e-09,
+		1.1781276e-09, 1.1962995e-09, 1.2158287e-09, 1.2369856e-09,
+		1.2601323e-09, 1.2857697e-09, 1.3146202e-09, 1.347784e-09,
+		1.3870636e-09, 1.4357403e-09, 1.5008659e-09, 1.6030948e-09,
+	};
+
+	@(static)
+	fn := [128]f32{
+		1.00000000,  0.9635997,   0.9362827,   0.9130436,   0.89228165,
+		0.87324303,  0.8555006,   0.8387836,   0.8229072,   0.8077383,
+		0.793177,    0.7791461,   0.7655842,   0.7524416,   0.73967725,
+		0.7272569,   0.7151515,   0.7033361,   0.69178915,  0.68049186,
+		0.6694277,   0.658582,    0.6479418,   0.63749546,  0.6272325,
+		0.6171434,   0.6072195,   0.5974532,   0.58783704,  0.5783647,
+		0.56903,     0.5598274,   0.5507518,   0.54179835,  0.5329627,
+		0.52424055,  0.5156282,   0.50712204,  0.49871865,  0.49041483,
+		0.48220766,  0.4740943,   0.46607214,  0.4581387,   0.45029163,
+		0.44252872,  0.43484783,  0.427247,    0.41972435,  0.41227803,
+		0.40490642,  0.39760786,  0.3903808,   0.3832238,   0.37613547,
+		0.36911446,  0.3621595,   0.35526937,  0.34844297,  0.34167916,
+		0.33497685,  0.3283351,   0.3217529,   0.3152294,   0.30876362,
+		0.30235484,  0.29600215,  0.28970486,  0.2834622,   0.2772735,
+		0.27113807,  0.2650553,   0.25902456,  0.2530453,   0.24711695,
+		0.241239,    0.23541094,  0.22963232,  0.2239027,   0.21822165,
+		0.21258877,  0.20700371,  0.20146611,  0.19597565,  0.19053204,
+		0.18513499,  0.17978427,  0.17447963,  0.1692209,   0.16400786,
+		0.15884037,  0.15371831,  0.14864157,  0.14361008,  0.13862377,
+		0.13368265,  0.12878671,  0.12393598,  0.119130544, 0.11437051,
+		0.10965602,  0.104987256, 0.10036444,  0.095787846, 0.0912578,
+		0.08677467,  0.0823389,   0.077950984, 0.073611505, 0.06932112,
+		0.06508058,  0.06089077,  0.056752663, 0.0526674,   0.048636295,
+		0.044660863, 0.040742867, 0.03688439,  0.033087887, 0.029356318,
+		0.025693292, 0.022103304, 0.018592102, 0.015167298, 0.011839478,
+		0.008624485, 0.005548995, 0.0026696292,
+	};
+
+
+	for {
+		j := i32(uint32(r));
+		i := j & 0x7f;
+		x := f64(j) * f64(wn[i]);
+		if u32(abs(j)) < kn[i] {
+			// 99% of the time this will be hit
+			return x;
+		}
+
+		if i == 0 {
+			for {
+				x = -math.log(float64(r)) * (1.0/ rn);
+				y := -math.log(float64(r));
+				if y+y >= x*x {
+					break;
+				}
+			}
+			return j > 0 ? rn + x : -rn - x;
+		}
+		if fn[i]+f32(float64(r))*(fn[i-1]-fn[i]) < f32(math.exp(-0.5*x*x)) {
+			return x;
+		}
+	}
+
+	return 0; // NOTE(bill): Will never be hit but this is here for sanity's sake
+}
+

core/math/rand/rand.odin

@@ -0,0 +1,130 @@
+package rand
+
+Rand :: struct {
+	state: u64,
+	inc:   u64,
+}
+
+
+@(private, static)
+_GLOBAL_SEED_DATA := 1234567890;
+@(private, static)
+global_rand := create(u64(uintptr(&_GLOBAL_SEED_DATA)));
+@(private, static)
+global_rand_ptr := &global_rand;
+
+set_global_seed :: proc(seed: u64) {
+	init(global_rand_ptr, seed);
+}
+
+create :: proc(seed: u64) -> Rand {
+	r: Rand;
+	init(&r, seed);
+	return r;
+}
+
+init :: proc(r: ^Rand, seed: u64) {
+	r.state = 0;
+	r.inc = (seed << 1) | 1;
+	_random(r);
+	r.state += seed;
+	_random(r);
+}
+
+_random :: proc(r: ^Rand) -> u32 {
+	old_state := r.state;
+	r.state = old_state * 6364136223846793005 + (r.inc|1);
+	xor_shifted := u32(((old_state>>18) ~ old_state) >> 27);
+	rot := u32(old_state >> 59);
+	return (xor_shifted >> rot) | (xor_shifted << ((-rot) & 31));
+}
+
+uint32 :: proc(r: ^Rand = global_rand_ptr) -> u32 { return _random(r); }
+
+uint64 :: proc(r: ^Rand = global_rand_ptr) -> u64 {
+	a := u64(_random(r));
+	b := u64(_random(r));
+	return (a<<32) | b;
+}
+
+int31 :: proc(r: ^Rand = global_rand_ptr) -> i32 { return i32(uint32(r) << 1 >> 1); }
+int63 :: proc(r: ^Rand = global_rand_ptr) -> i64 { return i64(uint64(r) << 1 >> 1); }
+
+int31_max :: proc(n: i32, r: ^Rand = global_rand_ptr) -> i32 {
+	if n <= 0 do panic("Invalid argument to int31_max");
+	if n&(n-1) == 0 {
+		return int31(r) & (n-1);
+	}
+	max := i32((1<<31) - 1 - (1<<31)&u32(n));
+	v := int31(r);
+	for v > max {
+		v = int31(r);
+	}
+	return v % n;
+}
+
+int63_max :: proc(n: i64, r: ^Rand = global_rand_ptr) -> i64 {
+	if n <= 0 do panic("Invalid argument to int63_max");
+	if n&(n-1) == 0 {
+		return int63(r) & (n-1);
+	}
+	max := i64((1<<63) - 1 - (1<<63)&u64(n));
+	v := int63(r);
+	for v > max {
+		v = int63(r);
+	}
+	return v % n;
+}
+
+int_max :: proc(n: int, r: ^Rand = global_rand_ptr) -> int {
+	if n <= 0 do panic("Invalid argument to int_max");
+	when size_of(int) == 4 {
+		return int(int31_max(i32(n), r));
+	} else {
+		return int(int63_max(i64(n), r));
+	}
+}
+
+float64 :: proc(r: ^Rand = global_rand_ptr) -> f64 { return f64(int63_max(1<<53, r)) / (1 << 53); }
+float32 :: proc(r: ^Rand = global_rand_ptr) -> f32 { return f32(float64(r)); }
+
+float64_range :: proc(lo, hi: f64, r: ^Rand = global_rand_ptr) -> f64 { return (hi-lo)*float64(r) + lo; }
+float32_range :: proc(lo, hi: f32, r: ^Rand = global_rand_ptr) -> f32 { return (hi-lo)*float32(r) + lo; }
+
+
+read :: proc(p: []byte, r: ^Rand = global_rand_ptr) -> (n: int) {
+	pos := i8(0);
+	val := i64(0);
+	for n = 0; n < len(p); n += 1 {
+		if pos == 0 {
+			val = int63(r);
+			pos = 7;
+		}
+		p[n] = byte(val);
+		val >>= 8;
+		pos -= 1;
+	}
+	return;
+}
+
+// perm returns a slice of n ints in a pseudo-random permutation of integers in the range [0, n)
+perm :: proc(n: int, r: ^Rand = global_rand_ptr) -> []int {
+	m := make([]int, n);
+	for i := 0; i < n; i += 1 {
+		j := int_max(i+1);
+		m[i] = m[j];
+		m[j] = i;
+	}
+	return m;
+}
+
+
+shuffle :: proc(array: $T/[]$E, r: ^Rand = global_rand_ptr) {
+	n := i64(len(array));
+	if n < 2 do return;
+
+	for i := i64(0); i < n; i += 1 {
+		j := int63_max(n, r);
+		array[i], array[j] = array[j], array[i];
+	}
+}

core/mem.odin

@@ -1,326 +0,0 @@
-#import "fmt.odin";
-#import "os.odin";
-
-swap :: proc(b: u16) -> u16 #foreign __llvm_core "llvm.bswap.i16";
-swap :: proc(b: u32) -> u32 #foreign __llvm_core "llvm.bswap.i32";
-swap :: proc(b: u64) -> u64 #foreign __llvm_core "llvm.bswap.i64";
-
-
-set :: proc(data: rawptr, value: i32, len: int) -> rawptr #link_name "__mem_set" {
-	llvm_memset_64bit :: proc(dst: rawptr, val: byte, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memset.p0i8.i64";
-	llvm_memset_64bit(data, cast(byte)value, len, 1, false);
-	return data;
-}
-
-zero :: proc(data: rawptr, len: int) -> rawptr #link_name "__mem_zero" {
-	return set(data, 0, len);
-}
-
-copy :: proc(dst, src: rawptr, len: int) -> rawptr #link_name "__mem_copy" {
-	// NOTE(bill): This _must_ be implemented like C's memmove
-	llvm_memmove_64bit :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memmove.p0i8.p0i8.i64";
-	llvm_memmove_64bit(dst, src, len, 1, false);
-	return dst;
-}
-
-copy_non_overlapping :: proc(dst, src: rawptr, len: int) -> rawptr #link_name "__mem_copy_non_overlapping" {
-	// NOTE(bill): This _must_ be implemented like C's memcpy
-	llvm_memcpy_64bit :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) #foreign __llvm_core "llvm.memcpy.p0i8.p0i8.i64";
-	llvm_memcpy_64bit(dst, src, len, 1, false);
-	return dst;
-}
-
-compare :: proc(a, b: []byte) -> int #link_name "__mem_compare" {
-	n := min(a.count, b.count);
-	for i in 0..<n {
-		match {
-		case a[i] < b[i]:
-			return -1;
-		case a[i] > b[i]:
-			return +1;
-		}
-	}
-	return 0;
-}
-
-
-
-kilobytes :: proc(x: int) -> int #inline { return          (x) * 1024; }
-megabytes :: proc(x: int) -> int #inline { return kilobytes(x) * 1024; }
-gigabytes :: proc(x: int) -> int #inline { return megabytes(x) * 1024; }
-terabytes :: proc(x: int) -> int #inline { return gigabytes(x) * 1024; }
-
-is_power_of_two :: proc(x: int) -> bool {
-	if x <= 0 {
-		return false;
-	}
-	return (x & (x-1)) == 0;
-}
-
-align_forward :: proc(ptr: rawptr, align: int) -> rawptr {
-	assert(is_power_of_two(align));
-
-	a := cast(uint)align;
-	p := cast(uint)ptr;
-	modulo := p & (a-1);
-	if modulo != 0 {
-		p += a - modulo;
-	}
-	return cast(rawptr)p;
-}
-
-
-
-Allocation_Header :: struct {
-	size: int,
-}
-
-allocation_header_fill :: proc(header: ^Allocation_Header, data: rawptr, size: int) {
-	header.size = size;
-	ptr := cast(^int)(header+1);
-
-	for i := 0; cast(rawptr)ptr < data; i += 1 {
-		(ptr+i)^ = -1;
-	}
-}
-allocation_header :: proc(data: rawptr) -> ^Allocation_Header {
-	if data == nil {
-		return nil;
-	}
-	p := cast(^int)data;
-	for (p-1)^ == -1 {
-		p = (p-1);
-	}
-	return cast(^Allocation_Header)p-1;
-}
-
-
-
-
-
-// Custom allocators
-Arena :: struct {
-	backing:    Allocator,
-	offset:     int,
-	memory:     []byte,
-	temp_count: int,
-}
-
-Arena_Temp_Memory :: struct {
-	arena:          ^Arena,
-	original_count: int,
-}
-
-
-
-
-
-init_arena_from_memory :: proc(using a: ^Arena, data: []byte) {
-	backing    = Allocator{};
-	memory     = data[:0];
-	temp_count = 0;
-}
-
-init_arena_from_context :: proc(using a: ^Arena, size: int) {
-	backing = context.allocator;
-	memory = new_slice(byte, size);
-	temp_count = 0;
-}
-
-free_arena :: proc(using a: ^Arena) {
-	if backing.procedure != nil {
-		push_allocator backing {
-			free(memory);
-			memory = nil;
-			offset = 0;
-		}
-	}
-}
-
-arena_allocator :: proc(arena: ^Arena) -> Allocator {
-	return Allocator{
-		procedure = arena_allocator_proc,
-		data = arena,
-	};
-}
-
-arena_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
-                          size, alignment: int,
-                          old_memory: rawptr, old_size: int, flags: u64) -> rawptr {
-	using Allocator_Mode;
-	arena := cast(^Arena)allocator_data;
-
-	match mode {
-	case ALLOC:
-		total_size := size + alignment;
-
-		if arena.offset + total_size > arena.memory.count {
-			fmt.fprintln(os.stderr, "Arena out of memory");
-			return nil;
-		}
-
-		#no_bounds_check end := ^arena.memory[arena.offset];
-
-		ptr := align_forward(end, alignment);
-		arena.offset += total_size;
-		return zero(ptr, size);
-
-	case FREE:
-		// NOTE(bill): Free all at once
-		// Use Arena_Temp_Memory if you want to free a block
-
-	case FREE_ALL:
-		arena.offset = 0;
-
-	case RESIZE:
-		return default_resize_align(old_memory, old_size, size, alignment);
-	}
-
-	return nil;
-}
-
-begin_arena_temp_memory :: proc(a: ^Arena) -> Arena_Temp_Memory {
-	tmp: Arena_Temp_Memory;
-	tmp.arena = a;
-	tmp.original_count = a.memory.count;
-	a.temp_count += 1;
-	return tmp;
-}
-
-end_arena_temp_memory :: proc(using tmp: Arena_Temp_Memory) {
-	assert(arena.memory.count >= original_count);
-	assert(arena.temp_count > 0);
-	arena.memory.count = original_count;
-	arena.temp_count -= 1;
-}
-
-
-
-
-
-
-
-align_of_type_info :: proc(type_info: ^Type_Info) -> int {
-	prev_pow2 :: proc(n: i64) -> i64 {
-		if n <= 0 {
-			return 0;
-		}
-		n |= n >> 1;
-		n |= n >> 2;
-		n |= n >> 4;
-		n |= n >> 8;
-		n |= n >> 16;
-		n |= n >> 32;
-		return n - (n >> 1);
-	}
-
-	WORD_SIZE :: size_of(int);
-	MAX_ALIGN :: size_of([vector 64]f64); // TODO(bill): Should these constants be builtin constants?
-	using Type_Info;
-	match info in type_info {
-	case Named:
-		return align_of_type_info(info.base);
-	case Integer:
-		return info.size;
-	case Float:
-		return info.size;
-	case String:
-		return WORD_SIZE;
-	case Boolean:
-		return 1;
-	case Any:
-		return WORD_SIZE;
-	case Pointer:
-		return WORD_SIZE;
-	case Procedure:
-		return WORD_SIZE;
-	case Array:
-		return align_of_type_info(info.elem);
-	case Dynamic_Array:
-		return WORD_SIZE;
-	case Slice:
-		return WORD_SIZE;
-	case Vector:
-		size := size_of_type_info(info.elem);
-		count := cast(int)max(prev_pow2(cast(i64)info.count), 1);
-		total := size * count;
-		return clamp(total, 1, MAX_ALIGN);
-	case Tuple:
-		return info.align;
-	case Struct:
-		return info.align;
-	case Union:
-		return info.align;
-	case Raw_Union:
-		return info.align;
-	case Enum:
-		return align_of_type_info(info.base);
-	case Map:
-		return align_of_type_info(info.generated_struct);
-	}
-
-	return 0;
-}
-
-align_formula :: proc(size, align: int) -> int {
-	result := size + align-1;
-	return result - result%align;
-}
-
-size_of_type_info :: proc(type_info: ^Type_Info) -> int {
-	WORD_SIZE :: size_of(int);
-	using Type_Info;
-	match info in type_info {
-	case Named:
-		return size_of_type_info(info.base);
-	case Integer:
-		return info.size;
-	case Float:
-		return info.size;
-	case String:
-		return 2*WORD_SIZE;
-	case Boolean:
-		return 1;
-	case Any:
-		return 2*WORD_SIZE;
-	case Pointer:
-		return WORD_SIZE;
-	case Procedure:
-		return WORD_SIZE;
-	case Array:
-		count := info.count;
-		if count == 0 {
-			return 0;
-		}
-		size      := size_of_type_info(info.elem);
-		align     := align_of_type_info(info.elem);
-		alignment := align_formula(size, align);
-		return alignment*(count-1) + size;
-	case Dynamic_Array:
-		return size_of(rawptr) + 2*size_of(int) + size_of(Allocator);
-	case Slice:
-		return 2*WORD_SIZE;
-	case Vector:
-		count := info.count;
-		if count == 0 {
-			return 0;
-		}
-		size      := size_of_type_info(info.elem);
-		align     := align_of_type_info(info.elem);
-		alignment := align_formula(size, align);
-		return alignment*(count-1) + size;
-	case Struct:
-		return info.size;
-	case Union:
-		return info.size;
-	case Raw_Union:
-		return info.size;
-	case Enum:
-		return size_of_type_info(info.base);
-	case Map:
-		return size_of_type_info(info.generated_struct);
-	}
-
-	return 0;
-}
-

core/mem/alloc.odin

@@ -0,0 +1,152 @@
+package mem
+
+import "core:runtime"
+
+DEFAULT_ALIGNMENT :: 2*align_of(rawptr);
+
+Allocator_Mode :: enum byte {
+	Alloc,
+	Free,
+	Free_All,
+	Resize,
+}
+
+Allocator_Proc :: #type proc(allocator_data: rawptr, mode: Allocator_Mode,
+	                         size, alignment: int,
+	                         old_memory: rawptr, old_size: int, flags: u64 = 0, location := #caller_location) -> rawptr;
+
+
+Allocator :: struct {
+	procedure: Allocator_Proc,
+	data:      rawptr,
+}
+
+
+
+alloc :: inline proc(size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, loc := #caller_location) -> rawptr {
+	if size == 0 do return nil;
+	if allocator.procedure == nil do return nil;
+	return allocator.procedure(allocator.data, Allocator_Mode.Alloc, size, alignment, nil, 0, 0, loc);
+}
+
+free :: inline proc(ptr: rawptr, allocator := context.allocator, loc := #caller_location) {
+	if ptr == nil do return;
+	if allocator.procedure == nil do return;
+	allocator.procedure(allocator.data, Allocator_Mode.Free, 0, 0, ptr, 0, 0, loc);
+}
+
+free_all :: inline proc(allocator := context.allocator, loc := #caller_location) {
+	if allocator.procedure != nil {
+		allocator.procedure(allocator.data, Allocator_Mode.Free_All, 0, 0, nil, 0, 0, loc);
+	}
+}
+
+resize :: inline proc(ptr: rawptr, old_size, new_size: int, alignment: int = DEFAULT_ALIGNMENT, allocator := context.allocator, loc := #caller_location) -> rawptr {
+	if allocator.procedure == nil {
+		return nil;
+	}
+	if new_size == 0 {
+		free(ptr, allocator, loc);
+		return nil;
+	} else if ptr == nil {
+		return allocator.procedure(allocator.data, Allocator_Mode.Alloc, new_size, alignment, nil, 0, 0, loc);
+	}
+	return allocator.procedure(allocator.data, Allocator_Mode.Resize, new_size, alignment, ptr, old_size, 0, loc);
+}
+
+
+delete_string :: proc(str: string, allocator := context.allocator, loc := #caller_location) {
+	free(raw_data(str), allocator, loc);
+}
+delete_cstring :: proc(str: cstring, allocator := context.allocator, loc := #caller_location) {
+	free((^byte)(str), allocator, loc);
+}
+delete_dynamic_array :: proc(array: $T/[dynamic]$E, loc := #caller_location) {
+	free(raw_data(array), array.allocator, loc);
+}
+delete_slice :: proc(array: $T/[]$E, allocator := context.allocator, loc := #caller_location) {
+	free(raw_data(array), allocator, loc);
+}
+delete_map :: proc(m: $T/map[$K]$V, loc := #caller_location) {
+	raw := transmute(Raw_Map)m;
+	delete_slice(raw.hashes);
+	free(raw.entries.data, raw.entries.allocator, loc);
+}
+
+
+delete :: proc{
+	delete_string,
+	delete_cstring,
+	delete_dynamic_array,
+	delete_slice,
+	delete_map,
+};
+
+
+new :: inline proc($T: typeid, allocator := context.allocator, loc := #caller_location) -> ^T {
+	ptr := (^T)(alloc(size_of(T), align_of(T), allocator, loc));
+	if ptr != nil do ptr^ = T{};
+	return ptr;
+}
+new_clone :: inline proc(data: $T, allocator := context.allocator, loc := #caller_location) -> ^T {
+	ptr := (^T)(alloc(size_of(T), align_of(T), allocator, loc));
+	if ptr != nil do ptr^ = data;
+	return ptr;
+}
+
+
+make_slice :: proc($T: typeid/[]$E, auto_cast len: int, allocator := context.allocator, loc := #caller_location) -> T {
+	runtime.make_slice_error_loc(loc, len);
+	data := alloc(size_of(E)*len, align_of(E), allocator, loc);
+	s := Raw_Slice{data, len};
+	return transmute(T)s;
+}
+make_dynamic_array :: proc($T: typeid/[dynamic]$E, allocator := context.allocator, loc := #caller_location) -> T {
+	return make_dynamic_array_len_cap(T, 0, 16, allocator, loc);
+}
+make_dynamic_array_len :: proc($T: typeid/[dynamic]$E, auto_cast len: int, allocator := context.allocator, loc := #caller_location) -> T {
+	return make_dynamic_array_len_cap(T, len, len, allocator, loc);
+}
+make_dynamic_array_len_cap :: proc($T: typeid/[dynamic]$E, auto_cast len: int, auto_cast cap: int, allocator := context.allocator, loc := #caller_location) -> T {
+	runtime.make_dynamic_array_error_loc(loc, len, cap);
+	data := alloc(size_of(E)*cap, align_of(E), allocator, loc);
+	s := Raw_Dynamic_Array{data, len, cap, allocator};
+	return transmute(T)s;
+}
+make_map :: proc($T: typeid/map[$K]$E, auto_cast cap: int = 16, allocator := context.allocator, loc := #caller_location) -> T {
+	runtime.make_map_expr_error_loc(loc, cap);
+	context.allocator = allocator;
+
+	m: T;
+	reserve_map(&m, cap);
+	return m;
+}
+
+make :: proc{
+	make_slice,
+	make_dynamic_array,
+	make_dynamic_array_len,
+	make_dynamic_array_len_cap,
+	make_map,
+};
+
+
+
+default_resize_align :: proc(old_memory: rawptr, old_size, new_size, alignment: int, allocator := context.allocator, loc := #caller_location) -> rawptr {
+	if old_memory == nil do return alloc(new_size, alignment, allocator, loc);
+
+	if new_size == 0 {
+		free(old_memory, allocator, loc);
+		return nil;
+	}
+
+	if new_size == old_size do return old_memory;
+
+	new_memory := alloc(new_size, alignment, allocator, loc);
+	if new_memory == nil do return nil;
+
+	copy(new_memory, old_memory, min(old_size, new_size));;
+	free(old_memory, allocator, loc);
+	return new_memory;
+}
+

core/mem/allocators.odin

@@ -0,0 +1,622 @@
+package mem
+
+
+
+nil_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                           size, alignment: int,
+                           old_memory: rawptr, old_size: int, flags: u64 = 0, loc := #caller_location) -> rawptr {
+	return nil;
+}
+
+nil_allocator :: proc() -> Allocator {
+	return Allocator{
+		procedure = nil_allocator_proc,
+		data = nil,
+	};
+}
+
+// Custom allocators
+
+Arena :: struct {
+	data:       []byte,
+	offset:     int,
+	peak_used:  int,
+	temp_count: int,
+}
+
+Arena_Temp_Memory :: struct {
+	arena:       ^Arena,
+	prev_offset: int,
+}
+
+
+init_arena :: proc(a: ^Arena, data: []byte) {
+	a.data       = data;
+	a.offset     = 0;
+	a.peak_used  = 0;
+	a.temp_count = 0;
+}
+
+arena_allocator :: proc(arena: ^Arena) -> Allocator {
+	return Allocator{
+		procedure = arena_allocator_proc,
+		data = arena,
+	};
+}
+
+arena_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                             size, alignment: int,
+                             old_memory: rawptr, old_size: int, flags: u64, location := #caller_location) -> rawptr {
+	arena := cast(^Arena)allocator_data;
+
+	switch mode {
+	case .Alloc:
+		total_size := size + alignment;
+
+		if arena.offset + total_size > len(arena.data) {
+			return nil;
+		}
+
+		#no_bounds_check end := &arena.data[len(arena.data)];
+
+		ptr := align_forward(end, uintptr(alignment));
+		arena.offset += total_size;
+		arena.peak_used = max(arena.peak_used, arena.offset);
+		return zero(ptr, size);
+
+	case .Free:
+		// NOTE(bill): Free all at once
+		// Use Arena_Temp_Memory if you want to free a block
+
+	case .Free_All:
+		arena.offset = 0;
+
+	case .Resize:
+		return default_resize_align(old_memory, old_size, size, alignment, arena_allocator(arena));
+	}
+
+	return nil;
+}
+
+begin_arena_temp_memory :: proc(a: ^Arena) -> Arena_Temp_Memory {
+	tmp: Arena_Temp_Memory;
+	tmp.arena = a;
+	tmp.prev_offset = a.offset;
+	a.temp_count += 1;
+	return tmp;
+}
+
+end_arena_temp_memory :: proc(using tmp: Arena_Temp_Memory) {
+	assert(arena.offset >= prev_offset);
+	assert(arena.temp_count > 0);
+	arena.offset = prev_offset;
+	arena.temp_count -= 1;
+}
+
+
+
+Scratch_Allocator :: struct {
+	data:     []byte,
+	curr_offset: int,
+	prev_offset: int,
+	backup_allocator: Allocator,
+	leaked_allocations: [dynamic]rawptr,
+}
+
+scratch_allocator_init :: proc(scratch: ^Scratch_Allocator, data: []byte, backup_allocator := context.allocator) {
+	scratch.data = data;
+	scratch.curr_offset = 0;
+	scratch.prev_offset = 0;
+	scratch.backup_allocator = backup_allocator;
+}
+
+scratch_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                               size, alignment: int,
+                               old_memory: rawptr, old_size: int, flags: u64 = 0, loc := #caller_location) -> rawptr {
+
+	scratch := (^Scratch_Allocator)(allocator_data);
+
+	if scratch.data == nil {
+		DEFAULT_SCRATCH_BACKING_SIZE :: 1<<22;
+		scratch_allocator_init(scratch, make([]byte, 1<<22));
+	}
+
+	switch mode {
+	case Allocator_Mode.Alloc:
+		switch {
+		case scratch.curr_offset+size <= len(scratch.data):
+			offset := align_forward_uintptr(uintptr(scratch.curr_offset), uintptr(alignment));
+			ptr := &scratch.data[offset];
+			zero(ptr, size);
+			scratch.prev_offset = int(offset);
+			scratch.curr_offset = int(offset) + size;
+			return ptr;
+		case size <= len(scratch.data):
+			offset := align_forward_uintptr(uintptr(0), uintptr(alignment));
+			ptr := &scratch.data[offset];
+			zero(ptr, size);
+			scratch.prev_offset = int(offset);
+			scratch.curr_offset = int(offset) + size;
+			return ptr;
+		}
+		// TODO(bill): Should leaks be notified about? Should probably use a logging system that is built into the context system
+		a := scratch.backup_allocator;
+		if a.procedure == nil {
+			a = context.allocator;
+			scratch.backup_allocator = a;
+		}
+
+		ptr := alloc(size, alignment, a, loc);
+		if scratch.leaked_allocations == nil {
+			scratch.leaked_allocations = make([dynamic]rawptr, a);
+		}
+		append(&scratch.leaked_allocations, ptr);
+
+		return ptr;
+
+	case Allocator_Mode.Free:
+		last_ptr := rawptr(&scratch.data[scratch.prev_offset]);
+		if old_memory == last_ptr {
+			full_size := scratch.curr_offset - scratch.prev_offset;
+			scratch.curr_offset = scratch.prev_offset;
+			zero(last_ptr, full_size);
+			return nil;
+		}
+		// NOTE(bill): It's scratch memory, don't worry about freeing
+
+	case Allocator_Mode.Free_All:
+		scratch.curr_offset = 0;
+		scratch.prev_offset = 0;
+		for ptr in scratch.leaked_allocations {
+			free(ptr, scratch.backup_allocator);
+		}
+		clear(&scratch.leaked_allocations);
+
+	case Allocator_Mode.Resize:
+		last_ptr := rawptr(&scratch.data[scratch.prev_offset]);
+		if old_memory == last_ptr && len(scratch.data)-scratch.prev_offset >= size {
+			scratch.curr_offset = scratch.prev_offset+size;
+			return old_memory;
+		}
+		return scratch_allocator_proc(allocator_data, Allocator_Mode.Alloc, size, alignment, old_memory, old_size, flags, loc);
+	}
+
+	return nil;
+}
+
+scratch_allocator :: proc(scratch: ^Scratch_Allocator) -> Allocator {
+	return Allocator{
+		procedure = scratch_allocator_proc,
+		data = scratch,
+	};
+}
+
+
+
+
+Stack_Allocation_Header :: struct {
+	prev_offset: int,
+	padding:     int,
+}
+
+// Stack is a stack-like allocator which has a strict memory freeing order
+Stack :: struct {
+	data: []byte,
+	prev_offset: int,
+	curr_offset: int,
+	peak_used: int,
+}
+
+init_stack :: proc(s: ^Stack, data: []byte) {
+	s.data = data;
+	s.prev_offset = 0;
+	s.curr_offset = 0;
+	s.peak_used = 0;
+}
+
+stack_allocator :: proc(stack: ^Stack) -> Allocator {
+	return Allocator{
+		procedure = stack_allocator_proc,
+		data = stack,
+	};
+}
+
+
+stack_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                             size, alignment: int,
+                             old_memory: rawptr, old_size: int, flags: u64, location := #caller_location) -> rawptr {
+	s := cast(^Stack)allocator_data;
+
+	if s.data == nil {
+		return nil;
+	}
+
+	raw_alloc :: proc(s: ^Stack, size, alignment: int) -> rawptr {
+		curr_addr := uintptr(&s.data[0]) + uintptr(s.curr_offset);
+		padding := calc_padding_with_header(curr_addr, uintptr(alignment), size_of(Stack_Allocation_Header));
+		if s.curr_offset + padding + size > len(s.data) {
+			return nil;
+		}
+		s.prev_offset = s.curr_offset;
+		s.curr_offset += padding;
+
+		next_addr := curr_addr + uintptr(padding);
+		header := (^Stack_Allocation_Header)(next_addr - size_of(Stack_Allocation_Header));
+		header.padding = auto_cast padding;
+		header.prev_offset = auto_cast s.prev_offset;
+
+		s.curr_offset += size;
+
+		s.peak_used = max(s.peak_used, s.curr_offset);
+
+		return zero(rawptr(next_addr), size);
+	}
+
+	switch mode {
+	case .Alloc:
+		return raw_alloc(s, size, alignment);
+	case .Free:
+		if old_memory == nil {
+			return nil;
+		}
+		start := uintptr(&s.data[0]);
+		end := start + uintptr(len(s.data));
+		curr_addr := uintptr(old_memory);
+
+		if !(start <= curr_addr && curr_addr < end) {
+			panic("Out of bounds memory address passed to stack allocator (free)");
+			return nil;
+		}
+
+		if curr_addr >= start+uintptr(s.curr_offset) {
+			// NOTE(bill): Allow double frees
+			return nil;
+		}
+
+		header := (^Stack_Allocation_Header)(curr_addr - size_of(Stack_Allocation_Header));
+		old_offset := int(curr_addr - uintptr(header.padding) - uintptr(&s.data[0]));
+
+		if old_offset != int(header.prev_offset) {
+			panic("Out of order stack allocator free");
+			return nil;
+		}
+
+		s.curr_offset = int(old_offset);
+		s.prev_offset = int(header.prev_offset);
+
+
+	case .Free_All:
+		s.prev_offset = 0;
+		s.curr_offset = 0;
+
+	case .Resize:
+		if old_memory == nil {
+			return raw_alloc(s, size, alignment);
+		}
+		if size == 0 {
+			return nil;
+		}
+
+		start := uintptr(&s.data[0]);
+		end := start + uintptr(len(s.data));
+		curr_addr := uintptr(old_memory);
+		if !(start <= curr_addr && curr_addr < end) {
+			panic("Out of bounds memory address passed to stack allocator (resize)");
+			return nil;
+		}
+
+		if curr_addr >= start+uintptr(s.curr_offset) {
+			// NOTE(bill): Allow double frees
+			return nil;
+		}
+
+		if old_size == size {
+			return old_memory;
+		}
+
+		header := (^Stack_Allocation_Header)(curr_addr - size_of(Stack_Allocation_Header));
+		old_offset := int(curr_addr - uintptr(header.padding) - uintptr(&s.data[0]));
+
+		if old_offset != int(header.prev_offset) {
+			ptr := raw_alloc(s, size, alignment);
+			copy(ptr, old_memory, min(old_size, size));
+			return ptr;
+		}
+
+		old_memory_size := uintptr(s.curr_offset) - (curr_addr - start);
+		assert(old_memory_size == uintptr(old_size));
+
+		diff := size - old_size;
+		s.curr_offset += diff; // works for smaller sizes too
+		if diff > 0 {
+			zero(rawptr(curr_addr + uintptr(diff)), diff);
+		}
+
+		return old_memory;
+	}
+
+	return nil;
+}
+
+
+
+
+
+
+
+Small_Stack_Allocation_Header :: struct {
+	padding: u8,
+}
+
+// Small_Stack is a stack-like allocator which uses the smallest possible header but at the cost of non-strict memory freeing order
+Small_Stack :: struct {
+	data: []byte,
+	offset: int,
+	peak_used: int,
+}
+
+init_small_stack :: proc(s: ^Small_Stack, data: []byte) {
+	s.data = data;
+	s.offset = 0;
+	s.peak_used = 0;
+}
+
+small_stack_allocator :: proc(stack: ^Small_Stack) -> Allocator {
+	return Allocator{
+		procedure = small_stack_allocator_proc,
+		data = stack,
+	};
+}
+
+small_stack_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                                   size, alignment: int,
+                                   old_memory: rawptr, old_size: int, flags: u64, location := #caller_location) -> rawptr {
+	s := cast(^Small_Stack)allocator_data;
+
+	if s.data == nil {
+		return nil;
+	}
+
+	alignment = clamp(alignment, 1, 8*size_of(Stack_Allocation_Header{}.padding)/2);
+
+	raw_alloc :: proc(s: ^Small_Stack, size, alignment: int) -> rawptr {
+		curr_addr := uintptr(&s.data[0]) + uintptr(s.offset);
+		padding := calc_padding_with_header(curr_addr, uintptr(alignment), size_of(Small_Stack_Allocation_Header));
+		if s.offset + padding + size > len(s.data) {
+			return nil;
+		}
+		s.offset += padding;
+
+		next_addr := curr_addr + uintptr(padding);
+		header := (^Small_Stack_Allocation_Header)(next_addr - size_of(Small_Stack_Allocation_Header));
+		header.padding = auto_cast padding;
+
+		s.offset += size;
+
+		s.peak_used = max(s.peak_used, s.offset);
+
+		return zero(rawptr(next_addr), size);
+	}
+
+	switch mode {
+	case .Alloc:
+		return raw_alloc(s, size, alignment);
+	case .Free:
+		if old_memory == nil {
+			return nil;
+		}
+		start := uintptr(&s.data[0]);
+		end := start + uintptr(len(s.data));
+		curr_addr := uintptr(old_memory);
+
+		if !(start <= curr_addr && curr_addr < end) {
+			panic("Out of bounds memory address passed to stack allocator (free)");
+			return nil;
+		}
+
+		if curr_addr >= start+uintptr(s.offset) {
+			// NOTE(bill): Allow double frees
+			return nil;
+		}
+
+		header := (^Small_Stack_Allocation_Header)(curr_addr - size_of(Small_Stack_Allocation_Header));
+		old_offset := int(curr_addr - uintptr(header.padding) - uintptr(&s.data[0]));
+
+		s.offset = int(old_offset);
+
+	case .Free_All:
+		s.offset = 0;
+
+	case .Resize:
+		if old_memory == nil {
+			return raw_alloc(s, size, alignment);
+		}
+		if size == 0 {
+			return nil;
+		}
+
+		start := uintptr(&s.data[0]);
+		end := start + uintptr(len(s.data));
+		curr_addr := uintptr(old_memory);
+		if !(start <= curr_addr && curr_addr < end) {
+			panic("Out of bounds memory address passed to stack allocator (resize)");
+			return nil;
+		}
+
+		if curr_addr >= start+uintptr(s.offset) {
+			// NOTE(bill): Treat as a double free
+			return nil;
+		}
+
+		if old_size == size {
+			return old_memory;
+		}
+
+		ptr := raw_alloc(s, size, alignment);
+		copy(ptr, old_memory, min(old_size, size));
+		return ptr;
+	}
+
+	return nil;
+}
+
+
+
+
+
+Dynamic_Pool :: struct {
+	block_size:    int,
+	out_band_size: int,
+	alignment:     int,
+
+	unused_blocks:        [dynamic]rawptr,
+	used_blocks:          [dynamic]rawptr,
+	out_band_allocations: [dynamic]rawptr,
+
+	current_block: rawptr,
+	current_pos:   rawptr,
+	bytes_left:    int,
+
+	block_allocator: Allocator,
+}
+
+
+DYNAMIC_POOL_BLOCK_SIZE_DEFAULT       :: 65536;
+DYNAMIC_POOL_OUT_OF_BAND_SIZE_DEFAULT :: 6554;
+
+
+
+dynamic_pool_allocator_proc :: proc(allocator_data: rawptr, mode: Allocator_Mode,
+                                    size, alignment: int,
+                                    old_memory: rawptr, old_size: int,
+                                    flags: u64 = 0, loc := #caller_location) -> rawptr {
+	pool := (^Dynamic_Pool)(allocator_data);
+
+	switch mode {
+	case Allocator_Mode.Alloc:
+		return dynamic_pool_alloc(pool, size);
+	case Allocator_Mode.Free:
+		panic("Allocator_Mode.Free is not supported for a pool");
+	case Allocator_Mode.Free_All:
+		dynamic_pool_free_all(pool);
+	case Allocator_Mode.Resize:
+		panic("Allocator_Mode.Resize is not supported for a pool");
+		if old_size >= size {
+			return old_memory;
+		}
+		ptr := dynamic_pool_alloc(pool, size);
+		copy(ptr, old_memory, old_size);
+		return ptr;
+	}
+	return nil;
+}
+
+
+dynamic_pool_allocator :: proc(pool: ^Dynamic_Pool) -> Allocator {
+	return Allocator{
+		procedure = dynamic_pool_allocator_proc,
+		data = pool,
+	};
+}
+
+dynamic_pool_init :: proc(pool: ^Dynamic_Pool,
+                          block_allocator := context.allocator,
+                          array_allocator := context.allocator,
+                          block_size := DYNAMIC_POOL_BLOCK_SIZE_DEFAULT,
+                          out_band_size := DYNAMIC_POOL_OUT_OF_BAND_SIZE_DEFAULT,
+                          alignment := 8) {
+	pool.block_size      = block_size;
+	pool.out_band_size   = out_band_size;
+	pool.alignment       = alignment;
+	pool.block_allocator = block_allocator;
+	pool.out_band_allocations.allocator = array_allocator;
+	pool.       unused_blocks.allocator = array_allocator;
+	pool.         used_blocks.allocator = array_allocator;
+}
+
+dynamic_pool_destroy :: proc(using pool: ^Dynamic_Pool) {
+	dynamic_pool_free_all(pool);
+	delete(unused_blocks);
+	delete(used_blocks);
+
+	zero(pool, size_of(pool^));
+}
+
+
+dynamic_pool_alloc :: proc(using pool: ^Dynamic_Pool, bytes: int) -> rawptr {
+	cycle_new_block :: proc(using pool: ^Dynamic_Pool) {
+		if block_allocator.procedure == nil {
+			panic("You must call pool_init on a Pool before using it");
+		}
+
+		if current_block != nil {
+			append(&used_blocks, current_block);
+		}
+
+		new_block: rawptr;
+		if len(unused_blocks) > 0 {
+			new_block = pop(&unused_blocks);
+		} else {
+			new_block = block_allocator.procedure(block_allocator.data, Allocator_Mode.Alloc,
+			                                      block_size, alignment,
+			                                      nil, 0);
+		}
+
+		bytes_left = block_size;
+		current_pos = new_block;
+		current_block = new_block;
+	}
+
+
+	extra := alignment - (bytes % alignment);
+	bytes += extra;
+	if bytes >= out_band_size {
+		assert(block_allocator.procedure != nil);
+		memory := block_allocator.procedure(block_allocator.data, Allocator_Mode.Alloc,
+			                                block_size, alignment,
+			                                nil, 0);
+		if memory != nil {
+			append(&out_band_allocations, (^byte)(memory));
+		}
+		return memory;
+	}
+
+	if bytes_left < bytes {
+		cycle_new_block(pool);
+		if current_block == nil {
+			return nil;
+		}
+	}
+
+	memory := current_pos;
+	current_pos = ptr_offset((^byte)(current_pos), bytes);
+	bytes_left -= bytes;
+	return memory;
+}
+
+
+dynamic_pool_reset :: proc(using pool: ^Dynamic_Pool) {
+	if current_block != nil {
+		append(&unused_blocks, current_block);
+		current_block = nil;
+	}
+
+	for block in used_blocks {
+		append(&unused_blocks, block);
+	}
+	clear(&used_blocks);
+
+	for a in out_band_allocations {
+		free(a, block_allocator);
+	}
+	clear(&out_band_allocations);
+}
+
+dynamic_pool_free_all :: proc(using pool: ^Dynamic_Pool) {
+	dynamic_pool_reset(pool);
+
+	for block in unused_blocks {
+		free(block, block_allocator);
+	}
+	clear(&unused_blocks);
+}

core/mem/mem.odin

@@ -0,0 +1,242 @@
+package mem
+
+foreign _ {
+	@(link_name = "llvm.bswap.i16") swap16 :: proc(b: u16) -> u16 ---;
+	@(link_name = "llvm.bswap.i32") swap32 :: proc(b: u32) -> u32 ---;
+	@(link_name = "llvm.bswap.i64") swap64 :: proc(b: u64) -> u64 ---;
+}
+swap :: proc{swap16, swap32, swap64};
+
+
+
+set :: proc "contextless" (data: rawptr, value: byte, len: int) -> rawptr {
+	if data == nil do return nil;
+	if len < 0 do return data;
+	foreign _ {
+		when size_of(rawptr) == 8 {
+			@(link_name="llvm.memset.p0i8.i64")
+			llvm_memset :: proc(dst: rawptr, val: byte, len: int, align: i32, is_volatile: bool) ---;
+		} else {
+			@(link_name="llvm.memset.p0i8.i32")
+			llvm_memset :: proc(dst: rawptr, val: byte, len: int, align: i32, is_volatile: bool) ---;
+		}
+	}
+	llvm_memset(data, byte(value), len, 1, false);
+	return data;
+}
+zero :: inline proc "contextless" (data: rawptr, len: int) -> rawptr {
+	return set(data, 0, len);
+}
+zero_item :: inline proc "contextless" (item: $P/^$T) {
+	set(item, 0, size_of(T));
+}
+zero_slice :: proc "contextless" (data: $T/[]$E) {
+	if n := len(data); n > 0 {
+		zero(&data[0], size_of(E)*n);
+	}
+}
+
+
+copy :: proc "contextless" (dst, src: rawptr, len: int) -> rawptr {
+	if src == nil do return dst;
+	// NOTE(bill): This _must_ be implemented like C's memmove
+	foreign _ {
+		when size_of(rawptr) == 8 {
+			@(link_name="llvm.memmove.p0i8.p0i8.i64")
+			llvm_memmove :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) ---;
+		} else {
+			@(link_name="llvm.memmove.p0i8.p0i8.i32")
+			llvm_memmove :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) ---;
+		}
+	}
+	llvm_memmove(dst, src, len, 1, false);
+	return dst;
+}
+copy_non_overlapping :: proc "contextless" (dst, src: rawptr, len: int) -> rawptr {
+	if src == nil do return dst;
+	// NOTE(bill): This _must_ be implemented like C's memcpy
+	foreign _ {
+		when size_of(rawptr) == 8 {
+			@(link_name="llvm.memcpy.p0i8.p0i8.i64")
+	 		llvm_memcpy :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) ---;
+		} else {
+			@(link_name="llvm.memcpy.p0i8.p0i8.i32")
+	 		llvm_memcpy :: proc(dst, src: rawptr, len: int, align: i32, is_volatile: bool) ---;
+		}
+	}
+	llvm_memcpy(dst, src, len, 1, false);
+	return dst;
+}
+compare :: inline proc "contextless" (a, b: []byte) -> int {
+	return compare_byte_ptrs(&a[0], &b[0], min(len(a), len(b)));
+}
+compare_byte_ptrs :: proc "contextless" (a, b: ^byte, n: int) -> int #no_bounds_check {
+	x := slice_ptr(a, n);
+	y := slice_ptr(b, n);
+
+	SU :: size_of(uintptr);
+	fast := n/SU + 1;
+	offset := (fast-1)*SU;
+	curr_block := 0;
+	if n < SU {
+		fast = 0;
+	}
+
+	la := slice_ptr((^uintptr)(a), fast);
+	lb := slice_ptr((^uintptr)(b), fast);
+
+	for /**/; curr_block < fast; curr_block += 1 {
+		if la[curr_block] ~ lb[curr_block] != 0 {
+			for pos := curr_block*SU; pos < n; pos += 1 {
+				if x[pos] ~ y[pos] != 0 {
+					return (int(x[pos]) - int(y[pos])) < 0 ? -1 : +1;
+				}
+			}
+		}
+	}
+
+	for /**/; offset < n; offset += 1 {
+		if x[offset] ~ y[offset] != 0 {
+			return (int(x[offset]) - int(y[offset])) < 0 ? -1 : +1;
+		}
+	}
+
+	return 0;
+}
+
+compare_ptrs :: inline proc "contextless" (a, b: rawptr, n: int) -> int {
+	return compare_byte_ptrs((^byte)(a), (^byte)(b), n);
+}
+
+ptr_offset :: inline proc "contextless" (ptr: $P/^$T, n: int) -> P {
+	new := int(uintptr(ptr)) + size_of(T)*n;
+	return P(uintptr(new));
+}
+
+ptr_sub :: inline proc "contextless" (a, b: $P/^$T) -> int {
+	return (int(uintptr(a)) - int(uintptr(b)))/size_of(T);
+}
+
+slice_ptr :: inline proc "contextless" (ptr: ^$T, len: int) -> []T {
+	assert(len >= 0);
+	slice := Raw_Slice{data = ptr, len = len};
+	return transmute([]T)slice;
+}
+
+slice_to_bytes :: inline proc "contextless" (slice: $E/[]$T) -> []byte {
+	s := transmute(Raw_Slice)slice;
+	s.len *= size_of(T);
+	return transmute([]byte)s;
+}
+
+
+buffer_from_slice :: inline proc(backing: $T/[]$E) -> [dynamic]E {
+	s := transmute(Raw_Slice)backing;
+	d := Raw_Dynamic_Array{
+		data      = s.data,
+		len       = 0,
+		cap       = s.len,
+		allocator = nil_allocator(),
+	};
+	return transmute([dynamic]E)d;
+}
+
+ptr_to_bytes :: inline proc "contextless" (ptr: ^$T, len := 1) -> []byte {
+	assert(len >= 0);
+	return transmute([]byte)Raw_Slice{ptr, len*size_of(T)};
+}
+
+any_to_bytes :: inline proc "contextless" (val: any) -> []byte {
+	ti := type_info_of(val.id);
+	size := ti != nil ? ti.size : 0;
+	return transmute([]byte)Raw_Slice{val.data, size};
+}
+
+
+kilobytes :: inline proc "contextless" (x: int) -> int do return          (x) * 1024;
+megabytes :: inline proc "contextless" (x: int) -> int do return kilobytes(x) * 1024;
+gigabytes :: inline proc "contextless" (x: int) -> int do return megabytes(x) * 1024;
+terabytes :: inline proc "contextless" (x: int) -> int do return gigabytes(x) * 1024;
+
+is_power_of_two :: inline proc(x: uintptr) -> bool {
+	if x <= 0 do return false;
+	return (x & (x-1)) == 0;
+}
+
+align_forward :: proc(ptr: rawptr, align: uintptr) -> rawptr {
+	assert(is_power_of_two(align));
+
+	a := uintptr(align);
+	p := uintptr(ptr);
+	modulo := p & (a-1);
+	if modulo != 0 do p += a - modulo;
+	return rawptr(p);
+}
+
+align_forward_uintptr :: proc(ptr, align: uintptr) -> uintptr {
+	assert(is_power_of_two(align));
+
+	a := uintptr(align);
+	p := uintptr(ptr);
+	modulo := p & (a-1);
+	if modulo != 0 do p += a - modulo;
+	return uintptr(p);
+}
+
+
+align_forward_int :: inline proc(ptr, align: int) -> int {
+	return int(align_forward_uintptr(uintptr(ptr), uintptr(align)));
+}
+align_forward_uint :: inline proc(ptr, align: uint) -> uint {
+	return uint(align_forward_uintptr(uintptr(ptr), uintptr(align)));
+}
+
+context_from_allocator :: proc(a: Allocator) -> type_of(context) {
+	context.allocator = a;
+	return context;
+}
+
+
+
+Fixed_Byte_Buffer :: distinct [dynamic]byte;
+
+make_fixed_byte_buffer :: proc(backing: []byte) -> Fixed_Byte_Buffer {
+	s := transmute(Raw_Slice)backing;
+	d: Raw_Dynamic_Array;
+	d.data = s.data;
+	d.len = 0;
+	d.cap = s.len;
+	d.allocator = nil_allocator();
+	return transmute(Fixed_Byte_Buffer)d;
+}
+
+
+
+align_formula :: proc(size, align: int) -> int {
+	result := size + align-1;
+	return result - result%align;
+}
+
+calc_padding_with_header :: proc(ptr: uintptr, align: uintptr, header_size: int) -> int {
+	p := uintptr(ptr);
+	a := uintptr(align);
+	modulo := p & (a-1);
+
+	padding := uintptr(0);
+	if modulo != 0 do padding = a - modulo;
+
+	needed_space := uintptr(header_size);
+	if padding < needed_space {
+		needed_space -= padding;
+
+		if needed_space & (a-1) > 0 {
+			padding += align * (1+(needed_space/align));
+		} else {
+			padding += align * (needed_space/align);
+		}
+	}
+
+	return int(padding);
+}
+
+

core/mem/raw.odin

@@ -0,0 +1,51 @@
+package mem
+
+Raw_Any :: struct {
+	data: rawptr,
+	id:   typeid,
+}
+
+Raw_String :: struct {
+	data: ^byte,
+	len:  int,
+}
+
+Raw_Cstring :: struct {
+	data: ^byte,
+}
+
+Raw_Slice :: struct {
+	data: rawptr,
+	len:  int,
+}
+
+Raw_Dynamic_Array :: struct {
+	data:      rawptr,
+	len:       int,
+	cap:       int,
+	allocator: Allocator,
+}
+
+Raw_Map :: struct {
+	hashes:  []int,
+	entries: Raw_Dynamic_Array,
+}
+
+
+make_any :: inline proc(data: rawptr, id: typeid) -> any {
+	return transmute(any)Raw_Any{data, id};
+}
+
+raw_string_data :: inline proc(s: $T/string) -> ^byte {
+	return (^Raw_String)(&s).data;
+}
+raw_slice_data :: inline proc(a: $T/[]$E) -> ^E {
+	return cast(^E)(^Raw_Slice)(&a).data;
+}
+raw_dynamic_array_data :: inline proc(a: $T/[dynamic]$E) -> ^E {
+	return cast(^E)(^Raw_Dynamic_Array)(&a).data;
+}
+
+raw_data :: proc{raw_string_data, raw_slice_data, raw_dynamic_array_data};
+
+

core/odin/ast/ast.odin

@@ -0,0 +1,616 @@
+package odin_ast
+
+import "core:odin/token"
+
+Proc_Tag :: enum {
+	Bounds_Check,
+	No_Bounds_Check,
+	Require_Results,
+}
+Proc_Tags :: distinct bit_set[Proc_Tag; u32];
+
+Proc_Inlining :: enum u32 {
+	None      = 0,
+	Inline    = 1,
+	No_Inline = 2,
+}
+
+Proc_Calling_Convention :: enum i32 {
+	Invalid = 0,
+	Odin,
+	Contextless,
+	C_Decl,
+	Std_Call,
+	Fast_Call,
+
+	Foreign_Block_Default = -1,
+}
+
+Node_State_Flag :: enum {
+	Bounds_Check,
+	No_Bounds_Check,
+}
+Node_State_Flags :: distinct bit_set[Node_State_Flag];
+
+
+Comment_Group :: struct {
+	list: []token.Token,
+}
+
+Node :: struct {
+	pos:         token.Pos,
+	end:         token.Pos,
+	derived:     any,
+	state_flags: Node_State_Flags,
+}
+
+
+Expr :: struct {
+	using expr_base: Node,
+}
+Stmt :: struct {
+	using stmt_base: Node,
+}
+Decl :: struct {
+	using decl_base: Stmt,
+}
+
+// Expressions
+
+Bad_Expr :: struct {
+	using node: Expr,
+}
+
+Ident :: struct {
+	using node: Expr,
+	name: string,
+}
+
+Implicit :: struct {
+	using node: Expr,
+	tok: token.Token,
+}
+
+
+Undef :: struct {
+	using node: Expr,
+	tok:  token.Kind,
+}
+
+Basic_Lit :: struct {
+	using node: Expr,
+	tok: token.Token,
+}
+
+Basic_Directive :: struct {
+	using node: Expr,
+	tok:  token.Token,
+	name: string,
+}
+
+Ellipsis :: struct {
+	using node: Expr,
+	tok:  token.Kind,
+	expr: ^Expr,
+}
+
+Proc_Lit :: struct {
+	using node: Expr,
+	type: ^Proc_Type,
+	body: ^Stmt,
+	tags: Proc_Tags,
+	inlining: Proc_Inlining,
+}
+
+Comp_Lit :: struct {
+	using node: Expr,
+	type: ^Expr,
+	open: token.Pos,
+	elems: []^Expr,
+	close: token.Pos,
+}
+
+
+Tag_Expr :: struct {
+	using node: Expr,
+	op:      token.Token,
+	name:    string,
+	expr:    ^Expr,
+}
+
+Unary_Expr :: struct {
+	using node: Expr,
+	op:   token.Token,
+	expr: ^Expr,
+}
+
+Binary_Expr :: struct {
+	using node: Expr,
+	left:  ^Expr,
+	op:    token.Token,
+	right: ^Expr,
+}
+
+Paren_Expr :: struct {
+	using node: Expr,
+	open:  token.Pos,
+	expr:  ^Expr,
+	close: token.Pos,
+}
+
+Selector_Expr :: struct {
+	using node: Expr,
+	expr:  ^Expr,
+	field: ^Ident,
+}
+
+Implicit_Selector_Expr :: struct {
+	using node: Expr,
+	field: ^Ident,
+}
+
+Index_Expr :: struct {
+	using node: Expr,
+	expr:  ^Expr,
+	open:  token.Pos,
+	index: ^Expr,
+	close: token.Pos,
+}
+
+Deref_Expr :: struct {
+	using node: Expr,
+	expr: ^Expr,
+	op:   token.Token,
+}
+
+Slice_Expr :: struct {
+	using node: Expr,
+	expr:     ^Expr,
+	open:     token.Pos,
+	low:      ^Expr,
+	interval: token.Token,
+	high:     ^Expr,
+	close:    token.Pos,
+}
+
+Call_Expr :: struct {
+	using node: Expr,
+	inlining: Proc_Inlining,
+	expr:     ^Expr,
+	open:     token.Pos,
+	args:     []^Expr,
+	ellipsis: token.Token,
+	close:    token.Pos,
+}
+
+Field_Value :: struct {
+	using node: Expr,
+	field: ^Expr,
+	sep:   token.Pos,
+	value: ^Expr,
+}
+
+Ternary_Expr :: struct {
+	using node: Expr,
+	cond: ^Expr,
+	op1:  token.Token,
+	x:    ^Expr,
+	op2:  token.Token,
+	y:    ^Expr,
+}
+
+Type_Assertion :: struct {
+	using node: Expr,
+	expr:  ^Expr,
+	dot:   token.Pos,
+	open:  token.Pos,
+	type:  ^Expr,
+	close: token.Pos,
+}
+
+Type_Cast :: struct {
+	using node: Expr,
+	tok:   token.Token,
+	open:  token.Pos,
+	type:  ^Expr,
+	close: token.Pos,
+	expr:  ^Expr,
+}
+
+Auto_Cast :: struct {
+	using node: Expr,
+	op:   token.Token,
+	expr: ^Expr,
+}
+
+
+
+
+// Statements
+
+Bad_Stmt :: struct {
+	using node: Stmt,
+}
+
+Empty_Stmt :: struct {
+	using node: Stmt,
+	semicolon: token.Pos, // Position of the following ';'
+}
+
+Expr_Stmt :: struct {
+	using node: Stmt,
+	expr: ^Expr,
+}
+
+Tag_Stmt :: struct {
+	using node: Stmt,
+	op:      token.Token,
+	name:    string,
+	stmt:    ^Stmt,
+}
+
+Assign_Stmt :: struct {
+	using node: Stmt,
+	lhs:    []^Expr,
+	op:     token.Token,
+	rhs:    []^Expr,
+}
+
+
+Block_Stmt :: struct {
+	using node: Stmt,
+	label: ^Expr,
+	open:  token.Pos,
+	stmts: []^Stmt,
+	close: token.Pos,
+}
+
+If_Stmt :: struct {
+	using node: Stmt,
+	label:     ^Expr,
+	if_pos:    token.Pos,
+	init:      ^Stmt,
+	cond:      ^Expr,
+	body:      ^Stmt,
+	else_stmt: ^Stmt,
+}
+
+When_Stmt :: struct {
+	using node: Stmt,
+	when_pos:  token.Pos,
+	cond:      ^Expr,
+	body:      ^Stmt,
+	else_stmt: ^Stmt,
+}
+
+Return_Stmt :: struct {
+	using node: Stmt,
+	results: []^Expr,
+}
+
+Defer_Stmt :: struct {
+	using node: Stmt,
+	stmt: ^Stmt,
+}
+
+For_Stmt :: struct {
+	using node: Stmt,
+	label:     ^Expr,
+	for_pos:   token.Pos,
+	init:      ^Stmt,
+	cond:      ^Expr,
+	post:      ^Stmt,
+	body:      ^Stmt,
+}
+
+Range_Stmt :: struct {
+	using node: Stmt,
+	label:     ^Expr,
+	for_pos:   token.Pos,
+	val0:      ^Expr,
+	val1:      ^Expr,
+	in_pos:    token.Pos,
+	expr:      ^Expr,
+	body:      ^Stmt,
+}
+
+
+Case_Clause :: struct {
+	using node: Stmt,
+	case_pos:   token.Pos,
+	list:       []^Expr,
+	terminator: token.Token,
+	body:       []^Stmt,
+}
+
+Switch_Stmt :: struct {
+	using node: Stmt,
+	label:      ^Expr,
+	switch_pos: token.Pos,
+	init:       ^Stmt,
+	cond:       ^Expr,
+	body:       ^Stmt,
+	complete:   bool,
+}
+
+Type_Switch_Stmt :: struct {
+	using node: Stmt,
+	label:      ^Expr,
+	switch_pos: token.Pos,
+	tag:        ^Stmt,
+	expr:       ^Expr,
+	body:       ^Stmt,
+	complete:   bool,
+}
+
+Branch_Stmt :: struct {
+	using node: Stmt,
+	tok:   token.Token,
+	label: ^Ident,
+}
+
+Using_Stmt :: struct {
+	using node: Stmt,
+	list: []^Expr,
+}
+
+
+// Declarations
+
+Bad_Decl :: struct {
+	using node: Decl,
+}
+
+Value_Decl :: struct {
+	using node: Decl,
+	docs:       ^Comment_Group,
+	attributes: [dynamic]^Attribute, // dynamic as parsing will add to them lazily
+	names:      []^Expr,
+	type:       ^Expr,
+	values:     []^Expr,
+	comment:    ^Comment_Group,
+	is_using:   bool,
+	is_mutable: bool,
+}
+
+Package_Decl :: struct {
+	using node: Decl,
+	docs:    ^Comment_Group,
+	token:   token.Token,
+	name:    string,
+	comment: ^Comment_Group,
+}
+
+Import_Decl :: struct {
+	using node: Decl,
+	docs:       ^Comment_Group,
+	is_using:    bool,
+	import_tok:  token.Token,
+	name:        token.Token,
+	relpath:     token.Token,
+	fullpath:    string,
+	comment:     ^Comment_Group,
+}
+
+Foreign_Block_Decl :: struct {
+	using node: Decl,
+	docs:            ^Comment_Group,
+	attributes:      [dynamic]^Attribute, // dynamic as parsing will add to them lazily
+	tok:             token.Token,
+	foreign_library: ^Expr,
+	body:            ^Stmt,
+}
+
+Foreign_Import_Decl :: struct {
+	using node: Decl,
+	docs:            ^Comment_Group,
+	foreign_tok:     token.Token,
+	import_tok:      token.Token,
+	name:            ^Ident,
+	collection_name: string,
+	fullpaths:       []string,
+	comment:         ^Comment_Group,
+}
+
+
+
+// Other things
+unparen_expr :: proc(expr: ^Expr) -> ^Expr {
+	if expr == nil {
+		return nil;
+	}
+	for {
+		e, ok := expr.derived.(Paren_Expr);
+		if !ok do break;
+		expr = e.expr;
+	}
+	return expr;
+}
+
+Field_Flag :: enum {
+	Ellipsis,
+	Using,
+	No_Alias,
+	C_Vararg,
+	Auto_Cast,
+	In,
+	Results,
+	Default_Parameters,
+	Typeid_Token,
+}
+
+Field_Flags :: distinct bit_set[Field_Flag];
+
+Field_Flags_Struct :: Field_Flags{
+	Field_Flag.Using,
+};
+Field_Flags_Record_Poly_Params :: Field_Flags{
+	Field_Flag.Typeid_Token,
+};
+Field_Flags_Signature :: Field_Flags{
+	Field_Flag.Ellipsis,
+	Field_Flag.Using,
+	Field_Flag.No_Alias,
+	Field_Flag.C_Vararg,
+	Field_Flag.Auto_Cast,
+	Field_Flag.Default_Parameters,
+};
+
+Field_Flags_Signature_Params  :: Field_Flags_Signature | {Field_Flag.Typeid_Token};
+Field_Flags_Signature_Results :: Field_Flags_Signature;
+
+
+Proc_Group :: struct {
+	using node: Expr,
+	tok:   token.Token,
+	open:  token.Pos,
+	args:  []^Expr,
+	close: token.Pos,
+}
+
+Attribute :: struct {
+	using node: Node,
+	tok:   token.Kind,
+	open:  token.Pos,
+	elems: []^Expr,
+	close: token.Pos,
+}
+
+Field :: struct {
+	using node: Node,
+	docs:          ^Comment_Group,
+	names:         []^Expr, // Could be polymorphic
+	type:          ^Expr,
+	default_value: ^Expr,
+	flags:         Field_Flags,
+	comment:       ^Comment_Group,
+}
+
+Field_List :: struct {
+	using node: Node,
+	open:  token.Pos,
+	list:  []^Field,
+	close: token.Pos,
+}
+
+
+// Types
+Typeid_Type :: struct {
+	using node: Expr,
+	tok:            token.Kind,
+	specialization: ^Expr,
+}
+
+Helper_Type :: struct {
+	using node: Expr,
+	tok:  token.Kind,
+	type: ^Expr,
+}
+
+Distinct_Type :: struct {
+	using node: Expr,
+	tok:  token.Kind,
+	type: ^Expr,
+}
+
+Opaque_Type :: struct {
+	using node: Expr,
+	tok:  token.Kind,
+	type: ^Expr,
+}
+
+Poly_Type :: struct {
+	using node: Expr,
+	dollar:         token.Pos,
+	type:           ^Ident,
+	specialization: ^Expr,
+}
+
+Proc_Type :: struct {
+	using node: Expr,
+	tok:       token.Token,
+	calling_convention: Proc_Calling_Convention,
+	params:    ^Field_List,
+	arrow:     token.Pos,
+	results:   ^Field_List,
+	tags:      Proc_Tags,
+	generic:   bool,
+	diverging: bool,
+}
+
+Pointer_Type :: struct {
+	using node: Expr,
+	pointer: token.Pos,
+	elem:    ^Expr,
+}
+
+Array_Type :: struct {
+	using node: Expr,
+	open:  token.Pos,
+	len:   ^Expr, // Ellipsis node for [?]T arrray types, nil for slice types
+	close: token.Pos,
+	elem:  ^Expr,
+}
+
+Dynamic_Array_Type :: struct {
+	using node: Expr,
+	open:        token.Pos,
+	dynamic_pos: token.Pos,
+	close:       token.Pos,
+	elem:        ^Expr,
+}
+
+Struct_Type :: struct {
+	using node: Expr,
+	tok_pos:   token.Pos,
+	poly_params:  ^Field_List,
+	align:        ^Expr,
+	is_packed:    bool,
+	is_raw_union: bool,
+	fields:       ^Field_List,
+	name_count:  int,
+}
+
+Union_Type :: struct {
+	using node: Expr,
+	tok_pos:     token.Pos,
+	poly_params: ^Field_List,
+	align:       ^Expr,
+	variants:    []^Expr,
+}
+
+Enum_Type :: struct {
+	using node: Expr,
+	tok_pos:  token.Pos,
+	base_type: ^Expr,
+	open:      token.Pos,
+	fields:    []^Expr,
+	close:     token.Pos,
+
+	is_using:  bool,
+}
+
+Bit_Field_Type :: struct {
+	using node: Expr,
+	tok_pos: token.Pos,
+	align:   ^Expr,
+	open:    token.Pos,
+	fields:  []^Field_Value, // Field_Value with ':' rather than '='
+	close:   token.Pos,
+}
+
+Bit_Set_Type :: struct {
+	using node: Expr,
+	tok_pos:    token.Pos,
+	open:       token.Pos,
+	elem:       ^Expr,
+	underlying: ^Expr,
+	close:      token.Pos,
+}
+
+Map_Type :: struct {
+	using node: Expr,
+	tok_pos: token.Pos,
+	key:     ^Expr,
+	value:   ^Expr,
+}

core/odin/ast/clone.odin

@@ -0,0 +1,314 @@
+package odin_ast
+
+import "core:mem"
+import "core:fmt"
+import "core:odin/token"
+
+new :: proc($T: typeid, pos, end: token.Pos) -> ^T {
+	n := mem.new(T);
+	n.pos = pos;
+	n.end = end;
+	n.derived = n^;
+	base: ^Node = n; // dummy check
+	_ = base; // "Use" type to make -vet happy
+	return n;
+}
+
+clone :: proc{
+	clone_node,
+	clone_expr,
+	clone_stmt,
+	clone_decl,
+	clone_array,
+	clone_dynamic_array,
+};
+
+clone_array :: proc(array: $A/[]^$T) -> A {
+	if len(array) == 0 {
+		return nil;
+	}
+	res := make(A, len(array));
+	for elem, i in array {
+		res[i] = auto_cast clone(elem);
+	}
+	return res;
+}
+
+clone_dynamic_array :: proc(array: $A/[dynamic]^$T) -> A {
+	if len(array) == 0 {
+		return nil;
+	}
+	res := make(A, len(array));
+	for elem, i in array {
+		res[i] = auto_cast clone(elem);
+	}
+	return res;
+}
+
+clone_expr :: proc(node: ^Expr) -> ^Expr {
+	return cast(^Expr)clone_node(node);
+}
+clone_stmt :: proc(node: ^Stmt) -> ^Stmt {
+	return cast(^Stmt)clone_node(node);
+}
+clone_decl :: proc(node: ^Decl) -> ^Decl {
+	return cast(^Decl)clone_node(node);
+}
+clone_node :: proc(node: ^Node) -> ^Node {
+	if node == nil {
+		return nil;
+	}
+
+	size := size_of(Node);
+	align := align_of(Node);
+	ti := type_info_of(node.derived.id);
+	if ti != nil {
+		size = ti.size;
+		align = ti.align;
+	}
+
+	res := cast(^Node)mem.alloc(size, align);
+	src: rawptr = node;
+	if node.derived != nil {
+		src = node.derived.data;
+	}
+	mem.copy(res, src, size);
+	res.derived.data = rawptr(res);
+
+	switch n in node.derived {
+	case Bad_Expr:
+	case Ident:
+	case Implicit:
+	case Undef:
+	case Basic_Lit:
+
+	case Ellipsis:
+		r := cast(^Ellipsis)res;
+		r.expr = clone(r.expr);
+	case Proc_Lit:
+		r := cast(^Proc_Lit)res;
+		r.type = auto_cast clone(r.type);
+		r.body = clone(r.body);
+	case Comp_Lit:
+		r := cast(^Comp_Lit)res;
+		r.type  = clone(r.type);
+		r.elems = clone(r.elems);
+
+	case Tag_Expr:
+		r := cast(^Tag_Expr)res;
+		r.expr = clone(r.expr);
+	case Unary_Expr:
+		r := cast(^Unary_Expr)res;
+		r.expr = clone(r.expr);
+	case Binary_Expr:
+		r := cast(^Binary_Expr)res;
+		r.left  = clone(r.left);
+		r.right = clone(r.right);
+	case Paren_Expr:
+		r := cast(^Paren_Expr)res;
+		r.expr = clone(r.expr);
+	case Selector_Expr:
+		r := cast(^Selector_Expr)res;
+		r.expr = clone(r.expr);
+		r.field = auto_cast clone(r.field);
+	case Index_Expr:
+		r := cast(^Index_Expr)res;
+		r.expr = clone(r.expr);
+		r.index = clone(r.index);
+	case Deref_Expr:
+		r := cast(^Deref_Expr)res;
+		r.expr = clone(r.expr);
+	case Slice_Expr:
+		r := cast(^Slice_Expr)res;
+		r.expr = clone(r.expr);
+		r.low  = clone(r.low);
+		r.high = clone(r.high);
+	case Call_Expr:
+		r := cast(^Call_Expr)res;
+		r.expr = clone(r.expr);
+		r.args = clone(r.args);
+	case Field_Value:
+		r := cast(^Field_Value)res;
+		r.field = clone(r.field);
+		r.value = clone(r.value);
+	case Ternary_Expr:
+		r := cast(^Ternary_Expr)res;
+		r.cond = clone(r.cond);
+		r.x    = clone(r.x);
+		r.y    = clone(r.y);
+	case Type_Assertion:
+		r := cast(^Type_Assertion)res;
+		r.expr = clone(r.expr);
+		r.type = clone(r.type);
+	case Type_Cast:
+		r := cast(^Type_Cast)res;
+		r.type = clone(r.type);
+		r.expr = clone(r.expr);
+	case Auto_Cast:
+		r := cast(^Auto_Cast)res;
+		r.expr = clone(r.expr);
+
+	case Bad_Stmt:
+	case Empty_Stmt:
+	case Expr_Stmt:
+		r := cast(^Expr_Stmt)res;
+		r.expr = clone(r.expr);
+	case Tag_Stmt:
+		r := cast(^Expr_Stmt)res;
+		r.expr = clone(r.expr);
+
+	case Assign_Stmt:
+		r := cast(^Assign_Stmt)res;
+		r.lhs = clone(r.lhs);
+		r.rhs = clone(r.rhs);
+	case Block_Stmt:
+		r := cast(^Block_Stmt)res;
+		r.label = auto_cast clone(r.label);
+		r.stmts = clone(r.stmts);
+	case If_Stmt:
+		r := cast(^If_Stmt)res;
+		r.label     = auto_cast clone(r.label);
+		r.init      = clone(r.init);
+		r.cond      = clone(r.cond);
+		r.body      = clone(r.body);
+		r.else_stmt = clone(r.else_stmt);
+	case When_Stmt:
+		r := cast(^When_Stmt)res;
+		r.cond      = clone(r.cond);
+		r.body      = clone(r.body);
+		r.else_stmt = clone(r.else_stmt);
+	case Return_Stmt:
+		r := cast(^Return_Stmt)res;
+		r.results = clone(r.results);
+	case Defer_Stmt:
+		r := cast(^Defer_Stmt)res;
+		r.stmt = clone(r.stmt);
+	case For_Stmt:
+		r := cast(^For_Stmt)res;
+		r.label = auto_cast clone(r.label);
+		r.init = clone(r.init);
+		r.cond = clone(r.cond);
+		r.post = clone(r.post);
+		r.body = clone(r.body);
+	case Range_Stmt:
+		r := cast(^Range_Stmt)res;
+		r.label = auto_cast clone(r.label);
+		r.val0 = clone(r.val0);
+		r.val1 = clone(r.val1);
+		r.expr = clone(r.expr);
+		r.body = clone(r.body);
+	case Case_Clause:
+		r := cast(^Case_Clause)res;
+		r.list = clone(r.list);
+		r.body = clone(r.body);
+	case Switch_Stmt:
+		r := cast(^Switch_Stmt)res;
+		r.label = auto_cast clone(r.label);
+		r.init = clone(r.init);
+		r.cond = clone(r.cond);
+		r.body = clone(r.body);
+	case Type_Switch_Stmt:
+		r := cast(^Type_Switch_Stmt)res;
+		r.label = auto_cast clone(r.label);
+		r.tag  = clone(r.tag);
+		r.expr = clone(r.expr);
+		r.body = clone(r.body);
+	case Branch_Stmt:
+		r := cast(^Branch_Stmt)res;
+		r.label = auto_cast clone(r.label);
+	case Using_Stmt:
+		r := cast(^Using_Stmt)res;
+		r.list = clone(r.list);
+	case Bad_Decl:
+	case Value_Decl:
+		r := cast(^Value_Decl)res;
+		r.attributes = clone(r.attributes);
+		r.names      = clone(r.names);
+		r.type       = clone(r.type);
+		r.values     = clone(r.values);
+	case Package_Decl:
+	case Import_Decl:
+	case Foreign_Block_Decl:
+		r := cast(^Foreign_Block_Decl)res;
+		r.attributes      = clone(r.attributes);
+		r.foreign_library = clone(r.foreign_library);
+		r.body            = clone(r.body);
+	case Foreign_Import_Decl:
+		r := cast(^Foreign_Import_Decl)res;
+		r.name = auto_cast clone(r.name);
+	case Proc_Group:
+		r := cast(^Proc_Group)res;
+		r.args = clone(r.args);
+	case Attribute:
+		r := cast(^Attribute)res;
+		r.elems = clone(r.elems);
+	case Field:
+		r := cast(^Field)res;
+		r.names         = clone(r.names);
+		r.type          = clone(r.type);
+		r.default_value = clone(r.default_value);
+	case Field_List:
+		r := cast(^Field_List)res;
+		r.list = clone(r.list);
+	case Typeid_Type:
+		r := cast(^Typeid_Type)res;
+		r.specialization = clone(r.specialization);
+	case Helper_Type:
+		r := cast(^Helper_Type)res;
+		r.type = clone(r.type);
+	case Distinct_Type:
+		r := cast(^Distinct_Type)res;
+		r.type = clone(r.type);
+	case Opaque_Type:
+		r := cast(^Opaque_Type)res;
+		r.type = clone(r.type);
+	case Poly_Type:
+		r := cast(^Poly_Type)res;
+		r.type = auto_cast clone(r.type);
+		r.specialization = clone(r.specialization);
+	case Proc_Type:
+		r := cast(^Proc_Type)res;
+		r.params  = auto_cast clone(r.params);
+		r.results = auto_cast clone(r.results);
+	case Pointer_Type:
+		r := cast(^Pointer_Type)res;
+		r.elem = clone(r.elem);
+	case Array_Type:
+		r := cast(^Array_Type)res;
+		r.len  = clone(r.len);
+		r.elem = clone(r.elem);
+	case Dynamic_Array_Type:
+		r := cast(^Dynamic_Array_Type)res;
+		r.elem = clone(r.elem);
+	case Struct_Type:
+		r := cast(^Struct_Type)res;
+		r.poly_params = auto_cast clone(r.poly_params);
+		r.align = clone(r.align);
+		r.fields = auto_cast clone(r.fields);
+	case Union_Type:
+		r := cast(^Union_Type)res;
+		r.poly_params = auto_cast clone(r.poly_params);
+		r.align = clone(r.align);
+		r.variants = clone(r.variants);
+	case Enum_Type:
+		r := cast(^Enum_Type)res;
+		r.base_type = clone(r.base_type);
+		r.fields = clone(r.fields);
+	case Bit_Field_Type:
+		r := cast(^Bit_Field_Type)res;
+		r.fields = clone(r.fields);
+	case Bit_Set_Type:
+		r := cast(^Bit_Set_Type)res;
+		r.elem = clone(r.elem);
+		r.underlying = clone(r.underlying);
+	case Map_Type:
+		r := cast(^Map_Type)res;
+		r.key = clone(r.key);
+		r.value = clone(r.value);
+
+	case:
+		fmt.panicf("Unhandled node kind: %T", n);
+	}
+
+	return res;
+}

core/odin/ast/file.odin

@@ -0,0 +1,40 @@
+package odin_ast
+
+import "core:odin/token"
+
+Package_Kind :: enum {
+	Normal,
+	Runtime,
+	Init,
+}
+
+Package :: struct {
+	kind:     Package_Kind,
+	id:       int,
+	name:     string,
+	fullpath: string,
+	files:    []^File,
+
+	user_data: rawptr,
+}
+
+File :: struct {
+	id: int,
+	pkg: ^Package,
+
+	fullpath: string,
+	src:      []byte,
+
+	pkg_decl:  ^Package_Decl,
+	pkg_token: token.Token,
+	pkg_name:  string,
+
+	decls:   [dynamic]^Stmt,
+	imports: [dynamic]^Import_Decl,
+	directive_count: int,
+
+	comments: [dynamic]^Comment_Group,
+
+	syntax_warning_count: int,
+	syntax_error_count:   int,
+}

core/odin/token/token.odin

@@ -0,0 +1,337 @@
+package odin_token
+
+import "core:strings"
+
+Token :: struct {
+	kind: Kind,
+	text: string,
+	pos:  Pos,
+}
+
+Pos :: struct {
+	file:   string,
+	offset: int, // starting at 0
+	line:   int, // starting at 1
+	column: int, // starting at 1
+}
+
+pos_compare :: proc(lhs, rhs: Pos) -> int {
+	if lhs.offset != rhs.offset {
+		return (lhs.offset < rhs.offset) ? -1 : +1;
+	}
+	if lhs.line != rhs.line {
+		return (lhs.line < rhs.line) ? -1 : +1;
+	}
+	if lhs.column != rhs.column {
+		return (lhs.column < rhs.column) ? -1 : +1;
+	}
+	return strings.compare(lhs.file, rhs.file);
+}
+
+using Kind :: enum u32 {
+	Invalid,
+	EOF,
+	Comment,
+
+	B_Literal_Begin,
+		Ident,
+		Integer,
+		Float,
+		Imag,
+		Rune,
+		String,
+	B_Literal_End,
+
+	B_Operator_Begin,
+		Eq,
+		Not,
+		Hash,
+		At,
+		Dollar,
+		Pointer,
+		Question,
+		Add,
+		Sub,
+		Mul,
+		Quo,
+		Mod,
+		Mod_Mod,
+		And,
+		Or,
+		Xor,
+		And_Not,
+		Shl,
+		Shr,
+
+		Cmp_And,
+		Cmp_Or,
+
+	B_Assign_Op_Begin,
+		Add_Eq,
+		Sub_Eq,
+		Mul_Eq,
+		Quo_Eq,
+		Mod_Eq,
+		Mod_Mod_Eq,
+		And_Eq,
+		Or_Eq,
+		Xor_Eq,
+		And_Not_Eq,
+		Shl_Eq,
+		Shr_Eq,
+		Cmp_And_Eq,
+		Cmp_Or_Eq,
+	B_Assign_Op_End,
+
+		Arrow_Right,
+		Arrow_Left,
+		Double_Arrow_Right,
+		Undef,
+
+	B_Comparison_Begin,
+		Cmp_Eq,
+		Not_Eq,
+		Lt,
+		Gt,
+		Lt_Eq,
+		Gt_Eq,
+	B_Comparison_End,
+
+		Open_Paren,
+		Close_Paren,
+		Open_Bracket,
+		Close_Bracket,
+		Open_Brace,
+		Close_Brace,
+		Colon,
+		Semicolon,
+		Period,
+		Comma,
+		Ellipsis,
+		Range_Half,
+		Back_Slash,
+	B_Operator_End,
+
+	B_Keyword_Begin,
+		Import,
+		Foreign,
+		Package,
+		Typeid,
+		When,
+		If,
+		Else,
+		For,
+		Switch,
+		In,
+		Notin,
+		Do,
+		Case,
+		Break,
+		Continue,
+		Fallthrough,
+		Defer,
+		Return,
+		Proc,
+		Macro,
+		Struct,
+		Union,
+		Enum,
+		Bit_Field,
+		Bit_Set,
+		Map,
+		Dynamic,
+		Auto_Cast,
+		Cast,
+		Transmute,
+		Distinct,
+		Opaque,
+		Using,
+		Inline,
+		No_Inline,
+		Context,
+		Size_Of,
+		Align_Of,
+		Offset_Of,
+		Type_Of,
+		Const,
+		Asm,
+		Yield,
+		Await,
+	B_Keyword_End,
+
+	COUNT,
+
+	B_Custom_Keyword_Begin = COUNT+1,
+	// ... Custom keywords
+};
+
+tokens := [Kind.COUNT]string {
+	"Invalid",
+	"EOF",
+	"Comment",
+
+	"",
+	"identifier",
+	"integer",
+	"float",
+	"imaginary",
+	"rune",
+	"string",
+	"",
+
+	"",
+	"=",
+	"!",
+	"#",
+	"@",
+	"$",
+	"^",
+	"?",
+	"+",
+	"-",
+	"*",
+	"/",
+	"%",
+	"%%",
+	"&",
+	"|",
+	"~",
+	"&~",
+	"<<",
+	">>",
+
+	"&&",
+	"||",
+
+	"",
+	"+=",
+	"-=",
+	"*=",
+	"/=",
+	"%=",
+	"%%=",
+	"&=",
+	"|=",
+	"~=",
+	"&~=",
+	"<<=",
+	">>=",
+	"&&=",
+	"||=",
+	"",
+
+	"->",
+	"<-",
+	"=>",
+	"---",
+
+	"",
+	"==",
+	"!=",
+	"<",
+	">",
+	"<=",
+	">=",
+	"",
+
+	"(",
+	")",
+	"[",
+	"]",
+	"{",
+	"}",
+	":",
+	";",
+	".",
+	",",
+	"..",
+	"..<",
+	"\\",
+	"",
+
+	"",
+	"import",
+	"foreign",
+	"package",
+	"typeid",
+	"when",
+	"if",
+	"else",
+	"for",
+	"switch",
+	"in",
+	"notin",
+	"do",
+	"case",
+	"break",
+	"continue",
+	"fallthrough",
+	"defer",
+	"return",
+	"proc",
+	"macro",
+	"struct",
+	"union",
+	"enum",
+	"bit_field",
+	"bit_set",
+	"map",
+	"dynamic",
+	"auto_cast",
+	"cast",
+	"transmute",
+	"distinct",
+	"opaque",
+	"using",
+	"inline",
+	"no_inline",
+	"context",
+	"size_of",
+	"align_of",
+	"offset_of",
+	"type_of",
+	"const",
+	"asm",
+	"yield",
+	"await",
+	"",
+};
+
+custom_keyword_tokens: []string;
+
+to_string :: proc(kind: Kind) -> string {
+	if Invalid <= kind && kind < COUNT {
+		return tokens[kind];
+	}
+	if B_Custom_Keyword_Begin < kind {
+		n := int(u16(kind)-u16(B_Custom_Keyword_Begin));
+		if n < len(custom_keyword_tokens) {
+			return custom_keyword_tokens[n];
+		}
+	}
+
+	return "Invalid";
+}
+
+is_literal  :: proc(kind: Kind) -> bool { return B_Literal_Begin  < kind && kind < B_Literal_End;  }
+is_operator :: proc(kind: Kind) -> bool {
+	switch kind {
+	case B_Operator_Begin..B_Operator_End:
+		return true;
+	case In, Notin:
+		return true;
+	}
+	return false;
+}
+is_assignment_operator :: proc(kind: Kind) -> bool {
+	return B_Assign_Op_Begin < kind && kind < B_Assign_Op_End || kind == Eq;
+}
+is_keyword :: proc(kind: Kind) -> bool {
+	switch {
+	case B_Keyword_Begin < kind && kind < B_Keyword_End:
+		return true;
+	case B_Custom_Keyword_Begin < kind:
+		return true;
+	}
+	return false;
+}

core/odin/tokenizer/tokenizer.odin

@@ -0,0 +1,621 @@
+package odin_tokenizer
+
+import "core:fmt"
+import "core:odin/token"
+import "core:unicode/utf8"
+
+Error_Handler :: #type proc(pos: token.Pos, fmt: string, args: ..any);
+
+Tokenizer :: struct {
+	// Immutable data
+	path: string,
+	src:  []byte,
+	err:  Error_Handler,
+
+	// Tokenizing state
+	ch:          rune,
+	offset:      int,
+	read_offset: int,
+	line_offset: int,
+	line_count:  int,
+
+	// Mutable data
+	error_count: int,
+}
+
+init :: proc(t: ^Tokenizer, src: []byte, path: string, err: Error_Handler = default_error_handler) {
+	t.src = src;
+	t.err = err;
+	t.ch = ' ';
+	t.offset = 0;
+	t.read_offset = 0;
+	t.line_offset = 0;
+	t.line_count = len(src) > 0 ? 1 : 0;
+	t.error_count = 0;
+	t.path = path;
+
+	advance_rune(t);
+	if t.ch == utf8.RUNE_BOM {
+		advance_rune(t);
+	}
+}
+
+@(private)
+offset_to_pos :: proc(t: ^Tokenizer, offset: int) -> token.Pos {
+	line := t.line_count;
+	column := offset - t.line_offset + 1;
+
+	return token.Pos {
+		file = t.path,
+		offset = offset,
+		line = line,
+		column = column,
+	};
+}
+
+default_error_handler :: proc(pos: token.Pos, msg: string, args: ..any) {
+	fmt.printf_err("%s(%d:%d) ", pos.file, pos.line, pos.column);
+	fmt.printf_err(msg, ..args);
+	fmt.printf_err("\n");
+}
+
+error :: proc(t: ^Tokenizer, offset: int, msg: string, args: ..any) {
+	pos := offset_to_pos(t, offset);
+	if t.err != nil {
+		t.err(pos, msg, ..args);
+	}
+	t.error_count += 1;
+}
+
+advance_rune :: proc(using t: ^Tokenizer) {
+	if read_offset < len(src) {
+		offset = read_offset;
+		if ch == '\n' {
+			line_offset = offset;
+			line_count += 1;
+		}
+		r, w := rune(src[read_offset]), 1;
+		switch {
+		case r == 0:
+			error(t, t.offset, "illegal character NUL");
+		case r >= utf8.RUNE_SELF:
+			r, w = utf8.decode_rune(src[read_offset:]);
+			if r == utf8.RUNE_ERROR && w == 1 {
+				error(t, t.offset, "illegal UTF-8 encoding");
+			} else if r == utf8.RUNE_BOM && offset > 0 {
+				error(t, t.offset, "illegal byte order mark");
+			}
+		}
+		read_offset += w;
+		ch = r;
+	} else {
+		offset = len(src);
+		if ch == '\n' {
+			line_offset = offset;
+			line_count += 1;
+		}
+		ch = -1;
+	}
+}
+
+peek_byte :: proc(using t: ^Tokenizer, offset := 0) -> byte {
+	if read_offset+offset < len(src) {
+		return src[read_offset+offset];
+	}
+	return 0;
+}
+
+skip_whitespace :: proc(t: ^Tokenizer) {
+	for t.ch == ' '  ||
+	    t.ch == '\t' ||
+	    t.ch == '\n' ||
+	    t.ch == '\r' {
+		advance_rune(t);
+	}
+}
+
+is_letter :: proc(r: rune) -> bool {
+	if r < utf8.RUNE_SELF {
+		switch r {
+		case '_':
+			return true;
+		case 'A'..'Z', 'a'..'z':
+			return true;
+		}
+	}
+	// TODO(bill): Add unicode lookup tables
+	return false;
+}
+is_digit :: proc(r: rune) -> bool {
+	// TODO(bill): Add unicode lookup tables
+	return '0' <= r && r <= '9';
+}
+
+
+scan_comment :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset-1;
+	next := -1;
+	general: {
+		if t.ch == '/' || t.ch == '!' { // // #! comments
+			advance_rune(t);
+			for t.ch != '\n' && t.ch >= 0 {
+				advance_rune(t);
+			}
+
+			next = t.offset;
+			if t.ch == '\n' {
+				next += 1;
+			}
+			break general;
+		}
+
+		/* style comment */
+		advance_rune(t);
+		for t.ch >= 0 {
+			ch := t.ch;
+			advance_rune(t);
+			if ch == '*' && t.ch == '/' {
+				advance_rune(t);
+				next = t.offset;
+				break general;
+			}
+		}
+
+		error(t, offset, "comment not terminated");
+	}
+
+	lit := t.src[offset : t.offset];
+
+	// NOTE(bill): Strip CR for line comments
+	for len(lit) > 2 && lit[1] == '/' && lit[len(lit)-1] == '\r' {
+		lit = lit[:len(lit)-1];
+	}
+
+
+	return string(lit);
+}
+
+scan_identifier :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset;
+
+	for is_letter(t.ch) || is_digit(t.ch) {
+		advance_rune(t);
+	}
+
+	return string(t.src[offset : t.offset]);
+}
+
+scan_string :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset-1;
+
+	for {
+		ch := t.ch;
+		if ch == '\n' || ch < 0 {
+			error(t, offset, "string literal was not terminated");
+			break;
+		}
+		advance_rune(t);
+		if ch == '"' {
+			break;
+		}
+		if ch == '\\' {
+			scan_escape(t);
+		}
+	}
+
+	return string(t.src[offset : t.offset]);
+}
+
+scan_raw_string :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset-1;
+
+	for {
+		ch := t.ch;
+		if ch == '\n' || ch < 0 {
+			error(t, offset, "raw string literal was not terminated");
+			break;
+		}
+		advance_rune(t);
+		if ch == '`' {
+			break;
+		}
+	}
+
+	return string(t.src[offset : t.offset]);
+}
+
+digit_val :: proc(r: rune) -> int {
+	switch r {
+	case '0'..'9':
+		return int(r-'0');
+	case 'A'..'F':
+		return int(r-'A' + 10);
+	case 'a'..'f':
+		return int(r-'a' + 10);
+	}
+	return 16;
+}
+
+scan_escape :: proc(t: ^Tokenizer) -> bool {
+	offset := t.offset;
+
+	n: int;
+	base, max: u32;
+	switch t.ch {
+	case 'a', 'b', 'e', 'f', 'n', 't', 'v', '\\', '\'', '\"':
+		advance_rune(t);
+		return true;
+
+	case '0'..'7':
+		n, base, max = 3, 8, 255;
+	case 'x':
+		advance_rune(t);
+		n, base, max = 2, 16, 255;
+	case 'u':
+		advance_rune(t);
+		n, base, max = 4, 16, utf8.MAX_RUNE;
+	case 'U':
+		advance_rune(t);
+		n, base, max = 8, 16, utf8.MAX_RUNE;
+	case:
+		if t.ch < 0 {
+			error(t, offset, "escape sequence was not terminated");
+		} else {
+			error(t, offset, "unknown escape sequence");
+		}
+		return false;
+	}
+
+	x: u32;
+	for n > 0 {
+		d := u32(digit_val(t.ch));
+		for d >= base {
+			if t.ch < 0 {
+				error(t, t.offset, "escape sequence was not terminated");
+			} else {
+				error(t, t.offset, "illegal character %d in escape sequence", t.ch);
+			}
+			return false;
+		}
+
+		x = x*base + d;
+		advance_rune(t);
+		n -= 1;
+	}
+
+	if x > max || 0xd800 <= x && x <= 0xe000 {
+		error(t, offset, "escape sequence is an invalid Unicode code point");
+		return false;
+	}
+	return true;
+}
+
+scan_rune :: proc(t: ^Tokenizer) -> string {
+	offset := t.offset-1;
+	valid := true;
+	n := 0;
+	for {
+		ch := t.ch;
+		if ch == '\n' || ch < 0 {
+			if valid {
+				error(t, offset, "rune literal not terminated");
+				valid = false;
+			}
+			break;
+		}
+		advance_rune(t);
+		if ch == '\'' {
+			break;
+		}
+		n += 1;
+		if ch == '\\' {
+			if !scan_escape(t)  {
+				valid = false;
+			}
+		}
+	}
+
+	if valid && n != 1 {
+		error(t, offset, "illegal rune literal");
+	}
+
+	return string(t.src[offset : t.offset]);
+}
+
+scan_number :: proc(t: ^Tokenizer, seen_decimal_point: bool) -> (token.Kind, string) {
+	scan_mantissa :: proc(t: ^Tokenizer, base: int) {
+		for digit_val(t.ch) < base || t.ch == '_' {
+			advance_rune(t);
+		}
+	}
+	scan_exponent :: proc(t: ^Tokenizer, kind: ^token.Kind) {
+		if t.ch == 'e' || t.ch == 'E' {
+			kind^ = token.Float;
+			advance_rune(t);
+			if t.ch == '-' || t.ch == '+' {
+				advance_rune(t);
+			}
+			if digit_val(t.ch) < 10 {
+				scan_mantissa(t, 10);
+			} else {
+				error(t, t.offset, "illegal floating-point exponent");
+			}
+		}
+
+		// NOTE(bill): This needs to be here for sanity's sake
+		if t.ch == 'i' {
+			kind^ = token.Imag;
+			advance_rune(t);
+		}
+	}
+	scan_fraction :: proc(t: ^Tokenizer, kind: ^token.Kind) -> (early_exit: bool) {
+		if t.ch == '.' && peek_byte(t) == '.' {
+			return true;
+		}
+		if t.ch == '.' {
+			kind^ = token.Float;
+			advance_rune(t);
+			scan_mantissa(t, 10);
+		}
+		return false;
+	}
+
+
+	offset := t.offset;
+	kind := token.Integer;
+
+	if seen_decimal_point {
+		offset -= 1;
+		kind = token.Float;
+		scan_mantissa(t, 10);
+		scan_exponent(t, &kind);
+	} else {
+		if t.ch == '0' {
+			int_base :: inline proc(t: ^Tokenizer, kind: ^token.Kind, base: int, msg: string) {
+				prev := t.offset;
+				advance_rune(t);
+				scan_mantissa(t, base);
+				if t.offset - prev <= 1 {
+					kind^ = token.Invalid;
+					error(t, t.offset, msg);
+				}
+			}
+
+			advance_rune(t);
+			switch t.ch {
+			case 'b': int_base(t, &kind,  2, "illegal binary integer");
+			case 'o': int_base(t, &kind,  8, "illegal octal integer");
+			case 'd': int_base(t, &kind, 10, "illegal decimal integer");
+			case 'z': int_base(t, &kind, 12, "illegal dozenal integer");
+			case 'x': int_base(t, &kind, 16, "illegal hexadecimal integer");
+			case 'h':
+				prev := t.offset;
+				advance_rune(t);
+				scan_mantissa(t, 16);
+				if t.offset - prev <= 1 {
+					kind = token.Invalid;
+					error(t, t.offset, "illegal hexadecimal floating-point number");
+				} else {
+					sub := t.src[prev+1 : t.offset];
+					digit_count := 0;
+					for d in sub {
+						if d != '_' {
+							digit_count += 1;
+						}
+					}
+
+					switch digit_count {
+					case 8, 16: break;
+					case:
+						error(t, t.offset, "invalid hexadecimal floating-point number, expected 8 or 16 digits, got %d", digit_count);
+					}
+				}
+
+			case:
+				seen_decimal_point = false;
+				scan_mantissa(t, 10);
+				if t.ch == '.' {
+					seen_decimal_point = true;
+					if scan_fraction(t, &kind) {
+						return kind, string(t.src[offset : t.offset]);
+					}
+				}
+				scan_exponent(t, &kind);
+				return kind, string(t.src[offset : t.offset]);
+			}
+		}
+	}
+
+	scan_mantissa(t, 10);
+
+	if scan_fraction(t, &kind) {
+		return kind, string(t.src[offset : t.offset]);
+	}
+
+	scan_exponent(t, &kind);
+
+	return kind, string(t.src[offset : t.offset]);
+}
+
+
+scan :: proc(t: ^Tokenizer) -> token.Token {
+	switch2 :: proc(t: ^Tokenizer, tok0, tok1: token.Kind) -> token.Kind {
+		if t.ch == '=' {
+			advance_rune(t);
+			return tok1;
+		}
+		return tok0;
+	}
+	switch3 :: proc(t: ^Tokenizer, tok0, tok1: token.Kind, ch2: rune, tok2: token.Kind) -> token.Kind {
+		if t.ch == '=' {
+			advance_rune(t);
+			return tok1;
+		}
+		if t.ch == ch2 {
+			advance_rune(t);
+			return tok2;
+		}
+		return tok0;
+	}
+	switch4 :: proc(t: ^Tokenizer, tok0, tok1: token.Kind, ch2: rune, tok2, tok3: token.Kind) -> token.Kind {
+		if t.ch == '=' {
+			advance_rune(t);
+			return tok1;
+		}
+		if t.ch == ch2 {
+			advance_rune(t);
+			if t.ch == '=' {
+				advance_rune(t);
+				return tok3;
+			}
+			return tok2;
+		}
+		return tok0;
+	}
+
+
+	skip_whitespace(t);
+
+	offset := t.offset;
+
+	kind: token.Kind;
+	lit:  string;
+	pos := offset_to_pos(t, offset);
+
+	switch ch := t.ch; true {
+	case is_letter(ch):
+		lit = scan_identifier(t);
+		kind = token.Ident;
+		check_keyword: if len(lit) > 1 {
+			// TODO(bill): Maybe have a hash table lookup rather than this linear search
+			for i in token.B_Keyword_Begin .. token.B_Keyword_End {
+				if lit == token.tokens[i] {
+					kind = token.Kind(i);
+					break check_keyword;
+				}
+			}
+			for keyword, i in token.custom_keyword_tokens {
+				if lit == keyword {
+					kind = token.Kind(i+1)+token.B_Custom_Keyword_Begin;
+					break check_keyword;
+				}
+			}
+		}
+	case '0' <= ch && ch <= '9':
+		kind, lit = scan_number(t, false);
+	case:
+		advance_rune(t);
+		switch ch {
+		case -1:
+			kind = token.EOF;
+		case '"':
+			kind = token.String;
+			lit = scan_string(t);
+		case '\'':
+			kind = token.Rune;
+			lit = scan_rune(t);
+		case '`':
+			kind = token.String;
+			lit = scan_raw_string(t);
+		case '=':
+			if t.ch == '>' {
+				advance_rune(t);
+				kind = token.Double_Arrow_Right;
+			} else {
+				kind = switch2(t, token.Eq, token.Cmp_Eq);
+			}
+		case '!': kind = switch2(t, token.Not, token.Not_Eq);
+		case '#':
+			kind = token.Hash;
+			if t.ch == '!' {
+				kind = token.Comment;
+				lit = scan_comment(t);
+			}
+		case '?': kind = token.Question;
+		case '@': kind = token.At;
+		case '$': kind = token.Dollar;
+		case '^': kind = token.Pointer;
+		case '+': kind = switch2(t, token.Add, token.Add_Eq);
+		case '-':
+			if t.ch == '>' {
+				advance_rune(t);
+				kind = token.Arrow_Right;
+			} else if t.ch == '-' && peek_byte(t) == '-' {
+				advance_rune(t);
+				advance_rune(t);
+				kind = token.Undef;
+			} else {
+				kind = switch2(t, token.Sub, token.Sub_Eq);
+			}
+		case '*': kind = switch2(t, token.Mul, token.Mul_Eq);
+		case '/':
+			if t.ch == '/' || t.ch == '*' {
+				kind = token.Comment;
+				lit = scan_comment(t);
+			} else {
+				kind = switch2(t, token.Quo, token.Quo_Eq);
+			}
+		case '%': kind = switch4(t, token.Mod, token.Mod_Eq, '%', token.Mod_Mod, token.Mod_Mod_Eq);
+		case '&':
+			if t.ch == '~' {
+				advance_rune(t);
+				kind = switch2(t, token.And_Not, token.And_Not_Eq);
+			} else {
+				kind = switch3(t, token.And, token.And_Eq, '&', token.Cmp_And);
+			}
+		case '|': kind = switch3(t, token.Or, token.Or_Eq, '|', token.Cmp_Or);
+		case '~': kind = token.Xor;
+		case '<':
+			if t.ch == '-' {
+				advance_rune(t);
+				kind = token.Arrow_Left;
+			} else {
+				kind = switch4(t, token.Lt, token.Lt_Eq, '<', token.Shl, token.Shl_Eq);
+			}
+		case '>': kind = switch4(t, token.Gt, token.Gt_Eq, '>', token.Shr,token.Shr_Eq);
+
+		case '≠': kind = token.Not_Eq;
+		case '≤': kind = token.Lt_Eq;
+		case '≥': kind = token.Gt_Eq;
+		case '∈': kind = token.In;
+		case '∉': kind = token.Notin;
+
+		case '.':
+			if '0' <= t.ch && t.ch <= '9' {
+				kind, lit = scan_number(t, true);
+			} else {
+				kind = token.Period;
+				if t.ch == '.' {
+					advance_rune(t);
+					kind = token.Ellipsis;
+					if t.ch == '<' {
+						advance_rune(t);
+						kind = token.Range_Half;
+					}
+				}
+			}
+		case ':': kind = token.Colon;
+		case ',': kind = token.Comma;
+		case ';': kind = token.Semicolon;
+		case '(': kind = token.Open_Paren;
+		case ')': kind = token.Close_Paren;
+		case '[': kind = token.Open_Bracket;
+		case ']': kind = token.Close_Bracket;
+		case '{': kind = token.Open_Brace;
+		case '}': kind = token.Close_Brace;
+
+		case '\\': kind = token.Back_Slash;
+
+		case:
+			if ch != utf8.RUNE_BOM {
+				error(t, t.offset, "illegal character '%r': %d", ch, ch);
+			}
+			kind = token.Invalid;
+		}
+	}
+
+	if lit == "" {
+		lit = string(t.src[offset : t.offset]);
+	}
+	return token.Token{kind, lit, pos};
+}

core/opengl.odin

@@ -1,155 +0,0 @@
-#foreign_system_library lib "opengl32.lib" when ODIN_OS == "windows";
-#import win32 "sys/windows.odin" when ODIN_OS == "windows";
-#load "opengl_constants.odin";
-
-Clear         :: proc(mask: u32)                                #foreign lib "glClear";
-ClearColor    :: proc(r, g, b, a: f32)                          #foreign lib "glClearColor";
-Begin         :: proc(mode: i32)                                #foreign lib "glBegin";
-End           :: proc()                                         #foreign lib "glEnd";
-Finish        :: proc()                                         #foreign lib "glFinish";
-BlendFunc     :: proc(sfactor, dfactor: i32)                    #foreign lib "glBlendFunc";
-Enable        :: proc(cap: i32)                                 #foreign lib "glEnable";
-Disable       :: proc(cap: i32)                                 #foreign lib "glDisable";
-GenTextures   :: proc(count: i32, result: ^u32)                 #foreign lib "glGenTextures";
-DeleteTextures:: proc(count: i32, result: ^u32)                 #foreign lib "glDeleteTextures";
-TexParameteri :: proc(target, pname, param: i32)                #foreign lib "glTexParameteri";
-TexParameterf :: proc(target: i32, pname: i32, param: f32)      #foreign lib "glTexParameterf";
-BindTexture   :: proc(target: i32, texture: u32)                #foreign lib "glBindTexture";
-LoadIdentity  :: proc()                                         #foreign lib "glLoadIdentity";
-Viewport      :: proc(x, y, width, height: i32)                 #foreign lib "glViewport";
-Ortho         :: proc(left, right, bottom, top, near, far: f64) #foreign lib "glOrtho";
-Color3f       :: proc(r, g, b: f32)                             #foreign lib "glColor3f";
-Vertex3f      :: proc(x, y, z: f32)                             #foreign lib "glVertex3f";
-TexImage2D    :: proc(target, level, internal_format,
-                      width, height, border,
-                      format, _type: i32, pixels: rawptr) #foreign lib "glTexImage2D";
-
-GetError    :: proc() -> i32            #foreign lib "glGetError";
-GetString   :: proc(name: i32) -> ^byte #foreign lib "glGetString";
-GetIntegerv :: proc(name: i32, v: ^i32) #foreign lib "glGetIntegerv";
-
-
-string_data :: proc(s: string) -> ^u8 #inline { return ^s[0]; }
-
-_libgl := win32.LoadLibraryA(string_data("opengl32.dll\x00"));
-
-GetProcAddress :: proc(name: string) -> proc() #cc_c {
-	assert(name[name.count-1] == 0);
-	res := win32.wglGetProcAddress(name.data);
-	if res == nil {
-		res = win32.GetProcAddress(_libgl, name.data);
-	}
-	return res;
-}
-
-GenBuffers:       proc(count: i32, buffers: ^u32) #cc_c;
-GenVertexArrays:  proc(count: i32, buffers: ^u32) #cc_c;
-GenSamplers:      proc(count: i32, buffers: ^u32) #cc_c;
-BindBuffer:       proc(target: i32, buffer: u32) #cc_c;
-BindVertexArray:  proc(buffer: u32) #cc_c;
-BindSampler:      proc(position: i32, sampler: u32) #cc_c;
-BufferData:       proc(target: i32, size: int, data: rawptr, usage: i32) #cc_c;
-BufferSubData:    proc(target: i32, offset, size: int, data: rawptr) #cc_c;
-
-DrawArrays:       proc(mode, first: i32, count: u32) #cc_c;
-DrawElements:     proc(mode: i32, count: u32, type_: i32, indices: rawptr) #cc_c;
-
-MapBuffer:        proc(target, access: i32) -> rawptr #cc_c;
-UnmapBuffer:      proc(target: i32) #cc_c;
-
-VertexAttribPointer:  proc(index: u32, size, type_: i32, normalized: i32, stride: u32, pointer: rawptr) #cc_c;
-EnableVertexAttribArray:  proc(index: u32) #cc_c;
-
-CreateShader:   proc(shader_type: i32) -> u32 #cc_c;
-ShaderSource:   proc(shader: u32, count: u32, str: ^^byte, length: ^i32) #cc_c;
-CompileShader:  proc(shader: u32) #cc_c;
-CreateProgram:  proc() -> u32 #cc_c;
-AttachShader:   proc(program, shader: u32) #cc_c;
-DetachShader:   proc(program, shader: u32) #cc_c;
-DeleteShader:   proc(shader:  u32) #cc_c;
-LinkProgram:    proc(program: u32) #cc_c;
-UseProgram:     proc(program: u32) #cc_c;
-DeleteProgram:  proc(program: u32) #cc_c;
-
-
-GetShaderiv:        proc(shader:  u32, pname: i32, params: ^i32) #cc_c;
-GetProgramiv:       proc(program: u32, pname: i32, params: ^i32) #cc_c;
-GetShaderInfoLog:   proc(shader:  u32, max_length: u32, length: ^u32, info_long: ^byte) #cc_c;
-GetProgramInfoLog:  proc(program: u32, max_length: u32, length: ^u32, info_long: ^byte) #cc_c;
-
-ActiveTexture:   proc(texture: i32) #cc_c;
-GenerateMipmap:  proc(target:  i32) #cc_c;
-
-SamplerParameteri:     proc(sampler: u32, pname: i32, param: i32) #cc_c;
-SamplerParameterf:     proc(sampler: u32, pname: i32, param: f32) #cc_c;
-SamplerParameteriv:    proc(sampler: u32, pname: i32, params: ^i32) #cc_c;
-SamplerParameterfv:    proc(sampler: u32, pname: i32, params: ^f32) #cc_c;
-SamplerParameterIiv:   proc(sampler: u32, pname: i32, params: ^i32) #cc_c;
-SamplerParameterIuiv:  proc(sampler: u32, pname: i32, params: ^u32) #cc_c;
-
-
-Uniform1i:         proc(loc: i32, v0: i32) #cc_c;
-Uniform2i:         proc(loc: i32, v0, v1: i32) #cc_c;
-Uniform3i:         proc(loc: i32, v0, v1, v2: i32) #cc_c;
-Uniform4i:         proc(loc: i32, v0, v1, v2, v3: i32) #cc_c;
-Uniform1f:         proc(loc: i32, v0: f32) #cc_c;
-Uniform2f:         proc(loc: i32, v0, v1: f32) #cc_c;
-Uniform3f:         proc(loc: i32, v0, v1, v2: f32) #cc_c;
-Uniform4f:         proc(loc: i32, v0, v1, v2, v3: f32) #cc_c;
-UniformMatrix4fv:  proc(loc: i32, count: u32, transpose: i32, value: ^f32) #cc_c;
-
-GetUniformLocation:  proc(program: u32, name: ^byte) -> i32 #cc_c;
-
-init :: proc() {
-	set_proc_address :: proc(p: rawptr, name: string) #inline { (cast(^(proc() #cc_c))p)^ = GetProcAddress(name); }
-
-	set_proc_address(^GenBuffers,      "glGenBuffers\x00");
-	set_proc_address(^GenVertexArrays, "glGenVertexArrays\x00");
-	set_proc_address(^GenSamplers,     "glGenSamplers\x00");
-	set_proc_address(^BindBuffer,      "glBindBuffer\x00");
-	set_proc_address(^BindSampler,     "glBindSampler\x00");
-	set_proc_address(^BindVertexArray, "glBindVertexArray\x00");
-	set_proc_address(^BufferData,      "glBufferData\x00");
-	set_proc_address(^BufferSubData,   "glBufferSubData\x00");
-
-	set_proc_address(^DrawArrays,      "glDrawArrays\x00");
-	set_proc_address(^DrawElements,    "glDrawElements\x00");
-
-	set_proc_address(^MapBuffer,   "glMapBuffer\x00");
-	set_proc_address(^UnmapBuffer, "glUnmapBuffer\x00");
-
-	set_proc_address(^VertexAttribPointer,     "glVertexAttribPointer\x00");
-	set_proc_address(^EnableVertexAttribArray, "glEnableVertexAttribArray\x00");
-
-	set_proc_address(^CreateShader,  "glCreateShader\x00");
-	set_proc_address(^ShaderSource,  "glShaderSource\x00");
-	set_proc_address(^CompileShader, "glCompileShader\x00");
-	set_proc_address(^CreateProgram, "glCreateProgram\x00");
-	set_proc_address(^AttachShader,  "glAttachShader\x00");
-	set_proc_address(^DetachShader,  "glDetachShader\x00");
-	set_proc_address(^DeleteShader,  "glDeleteShader\x00");
-	set_proc_address(^LinkProgram,   "glLinkProgram\x00");
-	set_proc_address(^UseProgram,    "glUseProgram\x00");
-	set_proc_address(^DeleteProgram, "glDeleteProgram\x00");
-
-	set_proc_address(^GetShaderiv,       "glGetShaderiv\x00");
-	set_proc_address(^GetProgramiv,      "glGetProgramiv\x00");
-	set_proc_address(^GetShaderInfoLog,  "glGetShaderInfoLog\x00");
-	set_proc_address(^GetProgramInfoLog, "glGetProgramInfoLog\x00");
-
-	set_proc_address(^ActiveTexture,  "glActiveTexture\x00");
-	set_proc_address(^GenerateMipmap, "glGenerateMipmap\x00");
-
-	set_proc_address(^Uniform1i,        "glUniform1i\x00");
-	set_proc_address(^UniformMatrix4fv, "glUniformMatrix4fv\x00");
-
-	set_proc_address(^GetUniformLocation, "glGetUniformLocation\x00");
-
-	set_proc_address(^SamplerParameteri,    "glSamplerParameteri\x00");
-	set_proc_address(^SamplerParameterf,    "glSamplerParameterf\x00");
-	set_proc_address(^SamplerParameteriv,   "glSamplerParameteriv\x00");
-	set_proc_address(^SamplerParameterfv,   "glSamplerParameterfv\x00");
-	set_proc_address(^SamplerParameterIiv,  "glSamplerParameterIiv\x00");
-	set_proc_address(^SamplerParameterIuiv, "glSamplerParameterIuiv\x00");
-}
-

core/os.odin

@@ -1,3 +0,0 @@
-#load "os_windows.odin" when ODIN_OS == "windows";
-#load "os_x.odin" when ODIN_OS == "osx";
-

core/os/os.odin

@@ -0,0 +1,154 @@
+package os
+
+import "core:mem"
+import "core:strconv"
+import "core:unicode/utf8"
+
+write_string :: proc(fd: Handle, str: string) -> (int, Errno) {
+	return write(fd, cast([]byte)str);
+}
+
+write_byte :: proc(fd: Handle, b: byte) -> (int, Errno) {
+	return write(fd, []byte{b});
+}
+
+write_rune :: proc(fd: Handle, r: rune) -> (int, Errno) {
+	if r < utf8.RUNE_SELF {
+		return write_byte(fd, byte(r));
+	}
+
+	b, n := utf8.encode_rune(r);
+	return write(fd, b[:n]);
+}
+
+write_encoded_rune :: proc(fd: Handle, r: rune) {
+	write_byte(fd, '\'');
+
+	switch r {
+	case '\a': write_string(fd, "\\a");
+	case '\b': write_string(fd, "\\b");
+	case '\e': write_string(fd, "\\e");
+	case '\f': write_string(fd, "\\f");
+	case '\n': write_string(fd, "\\n");
+	case '\r': write_string(fd, "\\r");
+	case '\t': write_string(fd, "\\t");
+	case '\v': write_string(fd, "\\v");
+	case:
+		if r < 32 {
+			write_string(fd, "\\x");
+			b: [2]byte;
+			s := strconv.append_bits(b[:], u64(r), 16, true, 64, strconv.digits, nil);
+			switch len(s) {
+			case 0: write_string(fd, "00");
+			case 1: write_rune(fd, '0');
+			case 2: write_string(fd, s);
+			}
+		} else {
+			write_rune(fd, r);
+		}
+	}
+	write_byte(fd, '\'');
+}
+
+
+file_size_from_path :: proc(path: string) -> i64 {
+	fd, err := open(path, O_RDONLY, 0);
+	if err != 0 {
+		return -1;
+	}
+	defer close(fd);
+
+	length: i64;
+	if length, err = file_size(fd); err != 0 {
+		return -1;
+	}
+	return length;
+}
+
+read_entire_file :: proc(name: string) -> (data: []byte, success: bool) {
+	fd, err := open(name, O_RDONLY, 0);
+	if err != 0 {
+		return nil, false;
+	}
+	defer close(fd);
+
+	length: i64;
+	if length, err = file_size(fd); err != 0 {
+		return nil, false;
+	}
+
+	if length <= 0 {
+		return nil, true;
+	}
+
+	data = make([]byte, int(length));
+	if data == nil {
+		return nil, false;
+	}
+
+	bytes_read, read_err := read(fd, data);
+	if read_err != 0 {
+		delete(data);
+		return nil, false;
+	}
+	return data[0:bytes_read], true;
+}
+
+write_entire_file :: proc(name: string, data: []byte, truncate := true) -> (success: bool) {
+	flags: int = O_WRONLY|O_CREATE;
+	if truncate {
+		flags |= O_TRUNC;
+	}
+	fd, err := open(name, flags, 0);
+	if err != 0 {
+		return false;
+	}
+	defer close(fd);
+
+	_, write_err := write(fd, data);
+	return write_err == 0;
+}
+
+write_ptr :: proc(fd: Handle, data: rawptr, len: int) -> (int, Errno) {
+	s := transmute([]byte)mem.Raw_Slice{data, len};
+	return write(fd, s);
+}
+
+read_ptr :: proc(fd: Handle, data: rawptr, len: int) -> (int, Errno) {
+	s := transmute([]byte)mem.Raw_Slice{data, len};
+	return read(fd, s);
+}
+
+
+heap_allocator_proc :: proc(allocator_data: rawptr, mode: mem.Allocator_Mode,
+                            size, alignment: int,
+                            old_memory: rawptr, old_size: int, flags: u64 = 0, loc := #caller_location) -> rawptr {
+	switch mode {
+	case .Alloc:
+		return heap_alloc(size);
+
+	case .Free:
+		heap_free(old_memory);
+		return nil;
+
+	case .Free_All:
+		// NOTE(bill): Does nothing
+
+	case .Resize:
+		if old_memory == nil {
+			return heap_alloc(size);
+		}
+		ptr := heap_resize(old_memory, size);
+		assert(ptr != nil);
+		return ptr;
+	}
+
+	return nil;
+}
+
+heap_allocator :: proc() -> mem.Allocator {
+	return mem.Allocator{
+		procedure = heap_allocator_proc,
+		data = nil,
+	};
+}

core/os/os_386.odin

@@ -0,0 +1,4 @@
+package os;
+
+ARCH :: "x86";
+ENDIAN :: "little";

core/os/os_amd64.odin

@@ -0,0 +1,4 @@
+package os;
+
+ARCH :: "amd64";
+ENDIAN :: "little";

core/os/os_essence.odin

@@ -0,0 +1,180 @@
+package os
+
+OS :: "essence";
+
+foreign import api "system:api"
+
+Handle    :: distinct int;
+Errno     :: distinct int;
+
+O_RDONLY   :: 0x00001;
+O_WRONLY   :: 0x00002;
+O_RDWR     :: 0x00003;
+O_CREATE   :: 0x00040;
+O_EXCL     :: 0x00080;
+O_TRUNC    :: 0x00200;
+O_APPEND   :: 0x00400;
+
+ERROR_NONE                           :: Errno(-1);
+ERROR_UNKNOWN_OPERATION_FAILURE      :: Errno(-7);
+ERROR_PATH_NOT_WITHIN_MOUNTED_VOLUME :: Errno(-14);
+ERROR_PATH_NOT_FOUND                 :: Errno(-15);
+ERROR_FILE_EXISTS                    :: Errno(-19);
+ERROR_FILE_NOT_FOUND                 :: Errno(-20);
+ERROR_DRIVE_ERROR_FILE_DAMAGED       :: Errno(-21);
+ERROR_ACCESS_NOT_WITHIN_FILE_BOUNDS  :: Errno(-22);
+ERROR_ACCESS_DENIED                  :: Errno(-23);
+ERROR_FILE_IN_EXCLUSIVE_USE          :: Errno(-24);
+ERROR_FILE_CANNOT_GET_EXCLUSIVE_USE  :: Errno(-25);
+ERROR_INCORRECT_NODE_TYPE            :: Errno(-26);
+ERROR_EVENT_NOT_SET                  :: Errno(-27);
+ERROR_TIMEOUT_REACHED                :: Errno(-29);
+ERROR_REQUEST_CLOSED_BEFORE_COMPLETE :: Errno(-30);
+ERROR_NO_CHARACTER_AT_COORDINATE     :: Errno(-31);
+ERROR_FILE_ON_READ_ONLY_VOLUME       :: Errno(-32);
+ERROR_USER_CANCELED_IO               :: Errno(-33);
+ERROR_DRIVE_CONTROLLER_REPORTED      :: Errno(-35);
+ERROR_COULD_NOT_ISSUE_PACKET         :: Errno(-36);
+
+ERROR_NOT_IMPLEMENTED                :: Errno(1);
+
+OS_Node_Type :: enum i32 {
+	File      = 0,
+	Directory = 1,
+}
+
+OS_Node_Information :: struct {
+	handle:   Handle,
+	id:       [16]byte,
+	ntype:    OS_Node_Type,
+	size:     i64,
+
+	// Our additions..
+	position: i64,
+}
+
+foreign api {
+	@(link_name="OSPrintDirect")    OSPrintDirect    :: proc(str: ^u8, length: int) ---;
+	@(link_name="malloc")           OSMalloc         :: proc(bytes: int) -> rawptr ---;
+	@(link_name="free")             OSFree           :: proc(address: rawptr) ---;
+	@(link_name="OSOpenNode")       OSOpenNode       :: proc(path: ^u8, path_length: int, flags: u64, information: ^OS_Node_Information) -> Errno ---;
+	@(link_name="OSResizeFile")     OSResizeFile     :: proc(handle: Handle, new_size: u64) -> Errno ---;
+	@(link_name="OSCloseHandle")    OSCloseHandle    :: proc(handle: Handle) ---;
+	@(link_name="OSWriteFileSync")  OSWriteFileSync  :: proc(handle: Handle, offset: i64, size: i64, buffer: rawptr) -> i64 ---;
+	@(link_name="OSReadFileSync")   OSReadFileSync   :: proc(handle: Handle, offset: i64, size: i64, buffer: rawptr) -> i64 ---;
+	@(link_name="realloc")          OSRealloc        :: proc(address: rawptr, size: int) -> rawptr ---;
+	@(link_name="OSGetThreadID")    OSGetThreadID    :: proc(handle: Handle) -> int ---;
+	@(link_name="OSRefreshNodeInformation") OSRefreshNodeInformation :: proc(information: ^OS_Node_Information) ---;
+}
+
+stdin  := Handle(-1); // Not implemented
+stdout := Handle(0);
+stderr := Handle(0);
+
+is_path_separator :: proc(r: rune) -> bool {
+	return r == '/';
+}
+
+current_thread_id :: proc "contextless" () -> int {
+	return OSGetThreadID(Handle(0x1000));
+}
+
+heap_alloc :: proc(size: int) -> rawptr {
+	return OSMalloc(size);
+}
+
+heap_free :: proc(address: rawptr) {
+	OSFree(address);
+}
+
+heap_resize :: proc(address: rawptr, new_size: int) -> rawptr {
+	return OSRealloc(address, new_size);
+}
+
+open :: proc(path: string, mode: int = O_RDONLY, perm: u32 = 0) -> (Handle, Errno) {
+	flags : u64 = 0;
+
+	if mode & O_CREATE == O_CREATE {
+		flags = flags | 0x9000; // Fail if found and create directories leading to the file if they don't exist
+	} else {
+		flags = flags | 0x2000; // Fail if not found
+	}
+
+	if mode & O_EXCL == O_EXCL {
+		flags = flags | 0x111; // Block opening the node for any reason
+	}
+
+	if mode & O_RDONLY == O_RDONLY {
+		flags = flags | 0x2; // Read access
+	}
+
+	if mode & O_WRONLY == O_WRONLY {
+		flags = flags | 0x220; // Write and resize access
+	}
+
+	if mode & O_TRUNC == O_TRUNC {
+		flags = flags | 0x200; // Resize access
+	}
+
+	information := new(OS_Node_Information);
+	error := OSOpenNode(&path[0], len(path), flags, information);
+
+	if error < ERROR_NONE {
+		free(information);
+		return 0, error;
+	}
+
+	if mode & O_TRUNC == O_TRUNC {
+		error := OSResizeFile(information.handle, 0);
+		if error < ERROR_NONE do return 0, ERROR_UNKNOWN_OPERATION_FAILURE;
+	}
+
+	if mode & O_APPEND == O_APPEND {
+		information.position = information.size;
+	} else {
+		information.position = 0;
+	}
+
+	return Handle(uintptr(information)), ERROR_NONE;
+}
+
+close :: proc(fd: Handle) {
+	information := (^OS_Node_Information)(uintptr(fd));
+	OSCloseHandle(information.handle);
+	free(information);
+}
+
+file_size :: proc(fd: Handle) -> (i64, Errno) {
+	x: OS_Node_Information;
+	OSRefreshNodeInformation(&x);
+	return x.size, ERROR_NONE;
+}
+
+write :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	if fd == 0 {
+		OSPrintDirect(&data[0], len(data));
+		return len(data), ERROR_NONE;
+	} else if fd == 1 {
+		assert(false);
+		return 0, ERROR_NOT_IMPLEMENTED;
+	}
+
+	information := (^OS_Node_Information)(uintptr(fd));
+	count := OSWriteFileSync(information.handle, information.position, i64(len(data)), &data[0]);
+	if count < 0 do  return 0, 1;
+	information.position += count;
+	return int(count), 0;
+}
+
+read :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	if (fd == 0 || fd == 1) {
+		assert(false);
+		return 0, ERROR_NOT_IMPLEMENTED;
+	}
+
+	information := (^OS_Node_Information)(uintptr(fd));
+	count := OSReadFileSync(information.handle, information.position, i64(len(data)), &data[0]);
+	if count < 0 do return 0, ERROR_UNKNOWN_OPERATION_FAILURE;
+	information.position += count;
+	return int(count), ERROR_NONE;
+}

core/os/os_linux.odin

@@ -0,0 +1,400 @@
+package os
+
+foreign import dl   "system:dl"
+foreign import libc "system:c"
+
+import "core:runtime"
+import "core:strings"
+
+OS :: "linux";
+
+Handle    :: distinct i32;
+File_Time :: distinct u64;
+Errno     :: distinct i32;
+Syscall   :: distinct int;
+
+INVALID_HANDLE :: ~Handle(0);
+
+ERROR_NONE:    Errno : 0;
+EPERM:         Errno : 1;
+ENOENT:        Errno : 2;
+EINTR:         Errno : 4;
+EIO:           Errno : 5;
+ENXIO:         Errno : 6;
+EBADF:         Errno : 9;
+EAGAIN:        Errno : 11;
+EWOULDBLOCK:   Errno : EAGAIN;
+ENOMEM:        Errno : 12;
+EACCES:        Errno : 13;
+EFAULT:        Errno : 14;
+EEXIST:        Errno : 17;
+ENODEV:        Errno : 19;
+ENOTDIR:       Errno : 20;
+EISDIR:        Errno : 21;
+EINVAL:        Errno : 22;
+ENFILE:        Errno : 23;
+EMFILE:        Errno : 24;
+ETXTBSY:       Errno : 26;
+EFBIG:         Errno : 27;
+ENOSPC:        Errno : 28;
+ESPIPE:        Errno : 29;
+EROFS:         Errno : 30;
+EPIPE:         Errno : 32;
+ENAMETOOLONG:  Errno : 36;
+ELOOP:         Errno : 40;
+EOVERFLOW:     Errno : 75;
+EDESTADDRREQ:  Errno : 89;
+EOPNOTSUPP:    Errno : 95;
+EDQUOT:        Errno : 122;
+
+O_RDONLY   :: 0x00000;
+O_WRONLY   :: 0x00001;
+O_RDWR     :: 0x00002;
+O_CREATE   :: 0x00040;
+O_EXCL     :: 0x00080;
+O_NOCTTY   :: 0x00100;
+O_TRUNC    :: 0x00200;
+O_NONBLOCK :: 0x00800;
+O_APPEND   :: 0x00400;
+O_SYNC     :: 0x01000;
+O_ASYNC    :: 0x02000;
+O_CLOEXEC  :: 0x80000;
+
+
+SEEK_SET   :: 0;
+SEEK_CUR   :: 1;
+SEEK_END   :: 2;
+SEEK_DATA  :: 3;
+SEEK_HOLE  :: 4;
+SEEK_MAX   :: SEEK_HOLE;
+
+// NOTE(zangent): These are OS specific!
+// Do not mix these up!
+RTLD_LAZY         :: 0x001;
+RTLD_NOW          :: 0x002;
+RTLD_BINDING_MASK :: 0x3;
+RTLD_GLOBAL       :: 0x100;
+
+// "Argv" arguments converted to Odin strings
+args := _alloc_command_line_arguments();
+
+_File_Time :: struct {
+	seconds:     i64,
+	nanoseconds: i64,
+}
+
+Stat :: struct {
+	device_id:     u64, // ID of device containing file
+	serial:        u64, // File serial number
+	nlink:         u64, // Number of hard links
+	mode:          u32, // Mode of the file
+	uid:           u32, // User ID of the file's owner
+	gid:           u32, // Group ID of the file's group
+	_padding:      i32, // 32 bits of padding
+	rdev:          u64, // Device ID, if device
+	size:          i64, // Size of the file, in bytes
+	block_size:    i64, // Optimal bllocksize for I/O
+	blocks:        i64, // Number of 512-byte blocks allocated
+
+	last_access:   _File_Time, // Time of last access
+	modified:      _File_Time, // Time of last modification
+	status_change: _File_Time, // Time of last status change
+
+	_reserve1,
+	_reserve2,
+	_reserve3:     i64,
+};
+
+// File type
+S_IFMT   :: 0o170000; // Type of file mask
+S_IFIFO  :: 0o010000; // Named pipe (fifo)
+S_IFCHR  :: 0o020000; // Character special
+S_IFDIR  :: 0o040000; // Directory
+S_IFBLK  :: 0o060000; // Block special
+S_IFREG  :: 0o100000; // Regular
+S_IFLNK  :: 0o120000; // Symbolic link
+S_IFSOCK :: 0o140000; // Socket
+
+// File mode
+// Read, write, execute/search by owner
+S_IRWXU :: 0o0700; // RWX mask for owner
+S_IRUSR :: 0o0400; // R for owner
+S_IWUSR :: 0o0200; // W for owner
+S_IXUSR :: 0o0100; // X for owner
+
+	// Read, write, execute/search by group
+S_IRWXG :: 0o0070; // RWX mask for group
+S_IRGRP :: 0o0040; // R for group
+S_IWGRP :: 0o0020; // W for group
+S_IXGRP :: 0o0010; // X for group
+
+	// Read, write, execute/search by others
+S_IRWXO :: 0o0007; // RWX mask for other
+S_IROTH :: 0o0004; // R for other
+S_IWOTH :: 0o0002; // W for other
+S_IXOTH :: 0o0001; // X for other
+
+S_ISUID :: 0o4000; // Set user id on execution
+S_ISGID :: 0o2000; // Set group id on execution
+S_ISVTX :: 0o1000; // Directory restrcted delete
+
+
+S_ISLNK  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFLNK;
+S_ISREG  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFREG;
+S_ISDIR  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFDIR;
+S_ISCHR  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFCHR;
+S_ISBLK  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFBLK;
+S_ISFIFO :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFIFO;
+S_ISSOCK :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFSOCK;
+
+F_OK :: 0; // Test for file existance
+X_OK :: 1; // Test for execute permission
+W_OK :: 2; // Test for write permission
+R_OK :: 4; // Test for read permission
+
+TimeSpec :: struct {
+	tv_sec  : i64,  /* seconds */
+	tv_nsec : i64,  /* nanoseconds */
+};
+
+CLOCK_REALTIME           :: 0;
+CLOCK_MONOTONIC          :: 1;
+CLOCK_PROCESS_CPUTIME_ID :: 2;
+CLOCK_THREAD_CPUTIME_ID  :: 3;
+CLOCK_MONOTONIC_RAW      :: 4;
+CLOCK_REALTIME_COARSE    :: 5;
+CLOCK_MONOTONIC_COARSE   :: 6;
+CLOCK_BOOTTIME           :: 7;
+CLOCK_REALTIME_ALARM     :: 8;
+CLOCK_BOOTTIME_ALARM     :: 9;
+
+SYS_GETTID: Syscall : 186;
+
+foreign libc {
+	@(link_name="__errno_location") __errno_location    :: proc() -> ^int ---;
+	@(link_name="syscall")          syscall             :: proc(number: Syscall, #c_vararg args: ..any) -> int ---;
+
+	@(link_name="open")             _unix_open          :: proc(path: cstring, flags: int, #c_vararg mode: ..any) -> Handle ---;
+	@(link_name="close")            _unix_close         :: proc(fd: Handle) -> int ---;
+	@(link_name="read")             _unix_read          :: proc(fd: Handle, buf: rawptr, size: int) -> int ---;
+	@(link_name="write")            _unix_write         :: proc(fd: Handle, buf: rawptr, size: int) -> int ---;
+	@(link_name="lseek64")          _unix_seek          :: proc(fd: Handle, offset: i64, whence: i32) -> i64 ---;
+	@(link_name="gettid")           _unix_gettid        :: proc() -> u64 ---;
+	@(link_name="stat")             _unix_stat          :: proc(path: cstring, stat: ^Stat) -> int ---;
+	@(link_name="fstat")            _unix_fstat         :: proc(fd: Handle, stat: ^Stat) -> int ---;
+	@(link_name="access")           _unix_access        :: proc(path: cstring, mask: int) -> int ---;
+
+	@(link_name="malloc")           _unix_malloc        :: proc(size: int) -> rawptr ---;
+	@(link_name="calloc")           _unix_calloc        :: proc(num, size: int) -> rawptr ---;
+	@(link_name="free")             _unix_free          :: proc(ptr: rawptr) ---;
+	@(link_name="realloc")          _unix_realloc       :: proc(ptr: rawptr, size: int) -> rawptr ---;
+	@(link_name="getenv")           _unix_getenv        :: proc(cstring) -> cstring ---;
+
+	@(link_name="clock_gettime")    _unix_clock_gettime :: proc(clock_id: u64, timespec: ^TimeSpec) ---;
+	@(link_name="nanosleep")        _unix_nanosleep     :: proc(requested: ^TimeSpec, remaining: ^TimeSpec) -> int ---;
+	@(link_name="sleep")            _unix_sleep         :: proc(seconds: u64) -> int ---;
+
+	@(link_name="exit")             _unix_exit          :: proc(status: int) -> ! ---;
+}
+foreign dl {
+	@(link_name="dlopen")           _unix_dlopen        :: proc(filename: cstring, flags: int) -> rawptr ---;
+	@(link_name="dlsym")            _unix_dlsym         :: proc(handle: rawptr, symbol: cstring) -> rawptr ---;
+	@(link_name="dlclose")          _unix_dlclose       :: proc(handle: rawptr) -> int ---;
+	@(link_name="dlerror")          _unix_dlerror       :: proc() -> cstring ---;
+}
+
+is_path_separator :: proc(r: rune) -> bool {
+	return r == '/';
+}
+
+get_last_error :: proc() -> int {
+	return __errno_location()^;
+}
+
+open :: proc(path: string, flags: int = O_RDONLY, mode: int = 0) -> (Handle, Errno) {
+	cstr := strings.clone_to_cstring(path);
+	handle := _unix_open(cstr, flags, mode);
+	delete(cstr);
+	if handle == -1 {
+		return INVALID_HANDLE, Errno(get_last_error());
+	}
+	return handle, ERROR_NONE;
+}
+
+close :: proc(fd: Handle) -> Errno {
+	result := _unix_close(fd);
+	if result == -1 {
+		return Errno(get_last_error());
+	}
+	return ERROR_NONE;
+}
+
+read :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	bytes_read := _unix_read(fd, &data[0], len(data));
+	if bytes_read == -1 {
+		return -1, Errno(get_last_error());
+	}
+	return bytes_read, ERROR_NONE;
+}
+
+write :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	bytes_written := _unix_write(fd, &data[0], len(data));
+	if bytes_written == -1 {
+		return -1, Errno(get_last_error());
+	}
+	return bytes_written, ERROR_NONE;
+}
+
+seek :: proc(fd: Handle, offset: i64, whence: int) -> (i64, Errno) {
+	res := _unix_seek(fd, offset, i32(whence));
+	if res == -1 {
+		return -1, Errno(get_last_error());
+	}
+	return res, ERROR_NONE;
+}
+
+file_size :: proc(fd: Handle) -> (i64, Errno) {
+	s, err := fstat(fd);
+	if err != ERROR_NONE {
+		return -1, err;
+	}
+	return s.size, ERROR_NONE;
+}
+
+
+// NOTE(bill): Uses startup to initialize it
+
+stdin:  Handle = 0;
+stdout: Handle = 1;
+stderr: Handle = 2;
+
+/* TODO(zangent): Implement these!
+last_write_time :: proc(fd: Handle) -> File_Time {}
+last_write_time_by_name :: proc(name: string) -> File_Time {}
+*/
+last_write_time :: proc(fd: Handle) -> (File_Time, Errno) {
+	s, err := fstat(fd);
+	if err != ERROR_NONE {
+		return 0, err;
+	}
+	return File_Time(s.modified.nanoseconds), ERROR_NONE;
+}
+
+last_write_time_by_name :: proc(name: string) -> (File_Time, Errno) {
+	s, err := stat(name);
+	if err != ERROR_NONE {
+		return 0, err;
+	}
+	return File_Time(s.modified.nanoseconds), ERROR_NONE;
+}
+
+stat :: inline proc(path: string) -> (Stat, Errno) {
+	cstr := strings.clone_to_cstring(path);
+	defer delete(cstr);
+
+	s: Stat;
+	result := _unix_stat(cstr, &s);
+	if result == -1 {
+		return s, Errno(get_last_error());
+	}
+	return s, ERROR_NONE;
+}
+
+fstat :: inline proc(fd: Handle) -> (Stat, Errno) {
+	s: Stat;
+	result := _unix_fstat(fd, &s);
+	if result == -1 {
+		return s, Errno(get_last_error());
+	}
+	return s, ERROR_NONE;
+}
+
+access :: inline proc(path: string, mask: int) -> (bool, Errno) {
+	cstr := strings.clone_to_cstring(path);
+	defer delete(cstr);
+	result := _unix_access(cstr, mask);
+	if result == -1 {
+		return false, Errno(get_last_error());
+	}
+	return true, ERROR_NONE;
+}
+
+heap_alloc :: proc(size: int) -> rawptr {
+	assert(size >= 0);
+	return _unix_calloc(1, size);
+}
+
+heap_resize :: proc(ptr: rawptr, new_size: int) -> rawptr {
+	return _unix_realloc(ptr, new_size);
+}
+
+heap_free :: proc(ptr: rawptr) {
+	_unix_free(ptr);
+}
+
+getenv :: proc(name: string) -> (string, bool) {
+	path_str := strings.clone_to_cstring(name);
+	defer delete(path_str);
+	cstr := _unix_getenv(path_str);
+	if cstr == nil {
+		return "", false;
+	}
+	return string(cstr), true;
+}
+
+exit :: proc(code: int) -> ! {
+	_unix_exit(code);
+}
+
+clock_gettime :: proc(clock_id: u64) -> TimeSpec {
+	ts : TimeSpec;
+	_unix_clock_gettime(clock_id, &ts);
+	return ts;
+}
+
+sleep :: proc(seconds: u64) -> int {
+
+	return _unix_sleep(seconds);
+}
+
+nanosleep :: proc(nanoseconds: i64) -> int {
+	assert(nanoseconds <= 999999999);
+	requested, remaining : TimeSpec;
+	requested = TimeSpec{tv_nsec = nanoseconds};
+
+	return _unix_nanosleep(&requested, &remaining);
+}
+
+current_thread_id :: proc "contextless" () -> int {
+	return syscall(SYS_GETTID);
+}
+
+dlopen :: inline proc(filename: string, flags: int) -> rawptr {
+	cstr := strings.clone_to_cstring(filename);
+	defer delete(cstr);
+	handle := _unix_dlopen(cstr, flags);
+	return handle;
+}
+dlsym :: inline proc(handle: rawptr, symbol: string) -> rawptr {
+	assert(handle != nil);
+	cstr := strings.clone_to_cstring(symbol);
+	defer delete(cstr);
+	proc_handle := _unix_dlsym(handle, cstr);
+	return proc_handle;
+}
+dlclose :: inline proc(handle: rawptr) -> bool {
+	assert(handle != nil);
+	return _unix_dlclose(handle) == 0;
+}
+dlerror :: proc() -> string {
+	return string(_unix_dlerror());
+}
+
+
+_alloc_command_line_arguments :: proc() -> []string {
+	args := make([]string, len(runtime.args__));
+	for arg, i in runtime.args__ {
+		args[i] = string(arg);
+	}
+	return args;
+}

core/os/os_osx.odin

@@ -0,0 +1,295 @@
+package os
+
+foreign import dl   "system:dl"
+foreign import libc "system:c"
+
+import "core:runtime"
+import "core:strings"
+
+OS :: "osx";
+
+Handle    :: distinct i32;
+File_Time :: distinct u64;
+Errno     :: distinct int;
+
+INVALID_HANDLE :: ~Handle(0);
+
+O_RDONLY   :: 0x00000;
+O_WRONLY   :: 0x00001;
+O_RDWR     :: 0x00002;
+O_CREATE   :: 0x00040;
+O_EXCL     :: 0x00080;
+O_NOCTTY   :: 0x00100;
+O_TRUNC    :: 0x00200;
+O_NONBLOCK :: 0x00800;
+O_APPEND   :: 0x00400;
+O_SYNC     :: 0x01000;
+O_ASYNC    :: 0x02000;
+O_CLOEXEC  :: 0x80000;
+
+
+SEEK_SET   :: 0;
+SEEK_CUR   :: 1;
+SEEK_END   :: 2;
+SEEK_DATA  :: 3;
+SEEK_HOLE  :: 4;
+SEEK_MAX   :: SEEK_HOLE;
+
+
+
+// NOTE(zangent): These are OS specific!
+// Do not mix these up!
+RTLD_LAZY     :: 0x1;
+RTLD_NOW      :: 0x2;
+RTLD_LOCAL    :: 0x4;
+RTLD_GLOBAL   :: 0x8;
+RTLD_NODELETE :: 0x80;
+RTLD_NOLOAD   :: 0x10;
+RTLD_FIRST    :: 0x100;
+
+
+// "Argv" arguments converted to Odin strings
+args := _alloc_command_line_arguments();
+
+_File_Time :: struct {
+	seconds: i64,
+	nanoseconds: i64,
+}
+
+Stat :: struct {
+	device_id:     i32, // ID of device containing file
+	mode:          u16, // Mode of the file
+	nlink:         u16, // Number of hard links
+	serial:        u64, // File serial number
+	uid:           u32, // User ID of the file's owner
+	gid:           u32, // Group ID of the file's group
+	rdev:          i32, // Device ID, if device
+
+	last_access:   File_Time, // Time of last access
+	modified:      File_Time, // Time of last modification
+	status_change: File_Time, // Time of last status change
+	created:       File_Time, // Time of creation
+
+	size:          i64,  // Size of the file, in bytes
+	blocks:        i64,  // Number of blocks allocated for the file
+	block_size:    i32,  // Optimal blocksize for I/O
+	flags:         u32,  // User-defined flags for the file
+	gen_num:       u32,  // File generation number ..?
+	_spare:        i32,  // RESERVED
+	_reserve1,
+	_reserve2:     i64,  // RESERVED
+};
+
+// File type
+S_IFMT   :: 0o170000; // Type of file mask
+S_IFIFO  :: 0o010000; // Named pipe (fifo)
+S_IFCHR  :: 0o020000; // Character special
+S_IFDIR  :: 0o040000; // Directory
+S_IFBLK  :: 0o060000; // Block special
+S_IFREG  :: 0o100000; // Regular
+S_IFLNK  :: 0o120000; // Symbolic link
+S_IFSOCK :: 0o140000; // Socket
+
+// File mode
+// Read, write, execute/search by owner
+S_IRWXU :: 0o0700; // RWX mask for owner
+S_IRUSR :: 0o0400; // R for owner
+S_IWUSR :: 0o0200; // W for owner
+S_IXUSR :: 0o0100; // X for owner
+
+// Read, write, execute/search by group
+S_IRWXG :: 0o0070; // RWX mask for group
+S_IRGRP :: 0o0040; // R for group
+S_IWGRP :: 0o0020; // W for group
+S_IXGRP :: 0o0010; // X for group
+
+// Read, write, execute/search by others
+S_IRWXO :: 0o0007; // RWX mask for other
+S_IROTH :: 0o0004; // R for other
+S_IWOTH :: 0o0002; // W for other
+S_IXOTH :: 0o0001; // X for other
+
+S_ISUID :: 0o4000; // Set user id on execution
+S_ISGID :: 0o2000; // Set group id on execution
+S_ISVTX :: 0o1000; // Directory restrcted delete
+
+S_ISLNK  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFLNK;
+S_ISREG  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFREG;
+S_ISDIR  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFDIR;
+S_ISCHR  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFCHR;
+S_ISBLK  :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFBLK;
+S_ISFIFO :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFIFO;
+S_ISSOCK :: inline proc(m: u32) -> bool do return (m & S_IFMT) == S_IFSOCK;
+
+R_OK :: 4; // Test for read permission
+W_OK :: 2; // Test for write permission
+X_OK :: 1; // Test for execute permission
+F_OK :: 0; // Test for file existance
+
+foreign libc {
+	@(link_name="open")    _unix_open    :: proc(path: cstring, flags: int, #c_vararg mode: ..any) -> Handle ---;
+	@(link_name="close")   _unix_close   :: proc(handle: Handle) ---;
+	@(link_name="read")    _unix_read    :: proc(handle: Handle, buffer: rawptr, count: int) -> int ---;
+	@(link_name="write")   _unix_write   :: proc(handle: Handle, buffer: rawptr, count: int) -> int ---;
+	@(link_name="lseek")   _unix_lseek   :: proc(fs: Handle, offset: int, whence: int) -> int ---;
+	@(link_name="gettid")  _unix_gettid  :: proc() -> u64 ---;
+	@(link_name="stat")    _unix_stat    :: proc(path: cstring, stat: ^Stat) -> int ---;
+	@(link_name="access")  _unix_access  :: proc(path: cstring, mask: int) -> int ---;
+
+	@(link_name="malloc")  _unix_malloc  :: proc(size: int) -> rawptr ---;
+	@(link_name="calloc")  _unix_calloc  :: proc(num, size: int) -> rawptr ---;
+	@(link_name="free")    _unix_free    :: proc(ptr: rawptr) ---;
+	@(link_name="realloc") _unix_realloc :: proc(ptr: rawptr, size: int) -> rawptr ---;
+	@(link_name="getenv")  _unix_getenv  :: proc(cstring) -> cstring ---;
+
+	@(link_name="exit")    _unix_exit    :: proc(status: int) ---;
+}
+
+foreign dl {
+	@(link_name="dlopen")  _unix_dlopen  :: proc(filename: cstring, flags: int) -> rawptr ---;
+	@(link_name="dlsym")   _unix_dlsym   :: proc(handle: rawptr, symbol: cstring) -> rawptr ---;
+	@(link_name="dlclose") _unix_dlclose :: proc(handle: rawptr) -> int ---;
+	@(link_name="dlerror") _unix_dlerror :: proc() -> cstring ---;
+}
+
+open :: proc(path: string, flags: int = O_RDONLY, mode: int = 0) -> (Handle, Errno) {
+	cstr := strings.clone_to_cstring(path);
+	handle := _unix_open(cstr, flags, mode);
+	delete(cstr);
+	if handle == -1 {
+		return INVALID_HANDLE, 1;
+	}
+	return handle, 0;
+}
+
+close :: proc(fd: Handle) {
+	_unix_close(fd);
+}
+
+write :: proc(fd: Handle, data: []u8) -> (int, Errno) {
+	assert(fd != -1);
+
+	bytes_written := _unix_write(fd, &data[0], len(data));
+	if(bytes_written == -1) {
+		return 0, 1;
+	}
+	return bytes_written, 0;
+}
+
+read :: proc(fd: Handle, data: []u8) -> (int, Errno) {
+	assert(fd != -1);
+
+	bytes_read := _unix_read(fd, &data[0], len(data));
+	if bytes_read == -1 {
+		return 0, 1;
+	}
+	return bytes_read, 0;
+}
+
+seek :: proc(fd: Handle, offset: i64, whence: int) -> (i64, Errno) {
+	assert(fd != -1);
+
+	final_offset := i64(_unix_lseek(fd, int(offset), whence));
+	if final_offset == -1 {
+		return 0, 1;
+	}
+	return final_offset, 0;
+}
+
+file_size :: proc(fd: Handle) -> (i64, Errno) {
+	prev, _   := seek(fd, 0, SEEK_CUR);
+	size, err := seek(fd, 0, SEEK_END);
+	seek(fd, prev, SEEK_SET);
+	return i64(size), err;
+}
+
+
+
+// NOTE(bill): Uses startup to initialize it
+stdin:  Handle = 0; // get_std_handle(win32.STD_INPUT_HANDLE);
+stdout: Handle = 1; // get_std_handle(win32.STD_OUTPUT_HANDLE);
+stderr: Handle = 2; // get_std_handle(win32.STD_ERROR_HANDLE);
+
+/* TODO(zangent): Implement these!
+last_write_time :: proc(fd: Handle) -> File_Time {}
+last_write_time_by_name :: proc(name: string) -> File_Time {}
+*/
+
+is_path_separator :: proc(r: rune) -> bool {
+	return r == '/';
+}
+
+stat :: inline proc(path: string) -> (Stat, bool) {
+	s: Stat;
+	cstr := strings.clone_to_cstring(path);
+	defer delete(cstr);
+	ret_int := _unix_stat(cstr, &s);
+	return s, ret_int==0;
+}
+
+access :: inline proc(path: string, mask: int) -> bool {
+	cstr := strings.clone_to_cstring(path);
+	defer delete(cstr);
+	return _unix_access(cstr, mask) == 0;
+}
+
+heap_alloc :: inline proc(size: int) -> rawptr {
+	assert(size > 0);
+	return _unix_calloc(1, size);
+}
+heap_resize :: inline proc(ptr: rawptr, new_size: int) -> rawptr {
+	return _unix_realloc(ptr, new_size);
+}
+heap_free :: inline proc(ptr: rawptr) {
+	_unix_free(ptr);
+}
+
+getenv :: proc(name: string) -> (string, bool) {
+	path_str := strings.clone_to_cstring(name);
+	defer delete(path_str);
+	cstr := _unix_getenv(path_str);
+	if cstr == nil {
+		return "", false;
+	}
+	return string(cstr), true;
+}
+
+exit :: inline proc(code: int) -> ! {
+	_unix_exit(code);
+}
+
+
+current_thread_id :: proc "contextless" () -> int {
+	// return int(_unix_gettid());
+	return 0;
+}
+
+dlopen :: inline proc(filename: string, flags: int) -> rawptr {
+	cstr := strings.clone_to_cstring(filename);
+	defer delete(cstr);
+	handle := _unix_dlopen(cstr, flags);
+	return handle;
+}
+dlsym :: inline proc(handle: rawptr, symbol: string) -> rawptr {
+	assert(handle != nil);
+	cstr := strings.clone_to_cstring(symbol);
+	defer delete(cstr);
+	proc_handle := _unix_dlsym(handle, cstr);
+	return proc_handle;
+}
+dlclose :: inline proc(handle: rawptr) -> bool {
+	assert(handle != nil);
+	return _unix_dlclose(handle) == 0;
+}
+dlerror :: proc() -> string {
+	return string(_unix_dlerror());
+}
+
+
+_alloc_command_line_arguments :: proc() -> []string {
+	args := make([]string, len(runtime.args__));
+	for arg, i in runtime.args__ {
+		args[i] = string(arg);
+	}
+	return args;
+}

core/os/os_windows.odin

@@ -0,0 +1,325 @@
+// +build windows
+package os
+
+import "core:sys/win32"
+
+OS :: "windows";
+
+Handle    :: distinct uintptr;
+File_Time :: distinct u64;
+Errno     :: distinct int;
+
+
+INVALID_HANDLE :: ~Handle(0);
+
+
+
+O_RDONLY   :: 0x00000;
+O_WRONLY   :: 0x00001;
+O_RDWR     :: 0x00002;
+O_CREATE   :: 0x00040;
+O_EXCL     :: 0x00080;
+O_NOCTTY   :: 0x00100;
+O_TRUNC    :: 0x00200;
+O_NONBLOCK :: 0x00800;
+O_APPEND   :: 0x00400;
+O_SYNC     :: 0x01000;
+O_ASYNC    :: 0x02000;
+O_CLOEXEC  :: 0x80000;
+
+
+ERROR_NONE:                   Errno : 0;
+ERROR_FILE_NOT_FOUND:         Errno : 2;
+ERROR_PATH_NOT_FOUND:         Errno : 3;
+ERROR_ACCESS_DENIED:          Errno : 5;
+ERROR_INVALID_HANDLE:         Errno : 6;
+ERROR_NO_MORE_FILES:          Errno : 18;
+ERROR_HANDLE_EOF:             Errno : 38;
+ERROR_NETNAME_DELETED:        Errno : 64;
+ERROR_FILE_EXISTS:            Errno : 80;
+ERROR_BROKEN_PIPE:            Errno : 109;
+ERROR_BUFFER_OVERFLOW:        Errno : 111;
+ERROR_INSUFFICIENT_BUFFER:    Errno : 122;
+ERROR_MOD_NOT_FOUND:          Errno : 126;
+ERROR_PROC_NOT_FOUND:         Errno : 127;
+ERROR_DIR_NOT_EMPTY:          Errno : 145;
+ERROR_ALREADY_EXISTS:         Errno : 183;
+ERROR_ENVVAR_NOT_FOUND:       Errno : 203;
+ERROR_MORE_DATA:              Errno : 234;
+ERROR_OPERATION_ABORTED:      Errno : 995;
+ERROR_IO_PENDING:             Errno : 997;
+ERROR_NOT_FOUND:              Errno : 1168;
+ERROR_PRIVILEGE_NOT_HELD:     Errno : 1314;
+WSAEACCES:                    Errno : 10013;
+WSAECONNRESET:                Errno : 10054;
+
+// Windows reserves errors >= 1<<29 for application use
+ERROR_FILE_IS_PIPE:           Errno : 1<<29 + 0;
+
+
+// "Argv" arguments converted to Odin strings
+args := _alloc_command_line_arguments();
+
+
+is_path_separator :: proc(r: rune) -> bool {
+	return r == '/' || r == '\\';
+}
+
+open :: proc(path: string, mode: int = O_RDONLY, perm: u32 = 0) -> (Handle, Errno) {
+	if len(path) == 0 do return INVALID_HANDLE, ERROR_FILE_NOT_FOUND;
+
+	access: u32;
+	switch mode & (O_RDONLY|O_WRONLY|O_RDWR) {
+	case O_RDONLY: access = win32.FILE_GENERIC_READ;
+	case O_WRONLY: access = win32.FILE_GENERIC_WRITE;
+	case O_RDWR:   access = win32.FILE_GENERIC_READ | win32.FILE_GENERIC_WRITE;
+	}
+
+	if mode&O_CREATE != 0 {
+		access |= win32.FILE_GENERIC_WRITE;
+	}
+	if mode&O_APPEND != 0 {
+		access &~= win32.FILE_GENERIC_WRITE;
+		access |=  win32.FILE_APPEND_DATA;
+	}
+
+	share_mode := u32(win32.FILE_SHARE_READ|win32.FILE_SHARE_WRITE);
+	sa: ^win32.Security_Attributes = nil;
+	sa_inherit := win32.Security_Attributes{length = size_of(win32.Security_Attributes), inherit_handle = true};
+	if mode&O_CLOEXEC == 0 {
+		sa = &sa_inherit;
+	}
+
+	create_mode: u32;
+	switch {
+	case mode&(O_CREATE|O_EXCL) == (O_CREATE | O_EXCL):
+		create_mode = win32.CREATE_NEW;
+	case mode&(O_CREATE|O_TRUNC) == (O_CREATE | O_TRUNC):
+		create_mode = win32.CREATE_ALWAYS;
+	case mode&O_CREATE == O_CREATE:
+		create_mode = win32.OPEN_ALWAYS;
+	case mode&O_TRUNC == O_TRUNC:
+		create_mode = win32.TRUNCATE_EXISTING;
+	case:
+		create_mode = win32.OPEN_EXISTING;
+	}
+	wide_path := win32.utf8_to_wstring(path);
+	handle := Handle(win32.create_file_w(wide_path, access, share_mode, sa, create_mode, win32.FILE_ATTRIBUTE_NORMAL, nil));
+	if handle != INVALID_HANDLE do return handle, ERROR_NONE;
+
+	err := Errno(win32.get_last_error());
+	return INVALID_HANDLE, err;
+}
+
+close :: proc(fd: Handle) -> Errno {
+	if win32.close_handle(win32.Handle(fd)) == 0 {
+		return Errno(win32.get_last_error());
+	}
+	return ERROR_NONE;
+}
+
+
+write :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	if len(data) == 0 do return 0, ERROR_NONE;
+
+	single_write_length: i32;
+	total_write: i64;
+	length := i64(len(data));
+
+	for total_write < length {
+		remaining := length - total_write;
+		MAX :: 1<<31-1;
+		to_write: i32 = min(i32(remaining), MAX);
+
+		e := win32.write_file(win32.Handle(fd), &data[total_write], to_write, &single_write_length, nil);
+		if single_write_length <= 0 || !e {
+			err := Errno(win32.get_last_error());
+			return int(total_write), err;
+		}
+		total_write += i64(single_write_length);
+	}
+	return int(total_write), ERROR_NONE;
+}
+
+read :: proc(fd: Handle, data: []byte) -> (int, Errno) {
+	if len(data) == 0 do return 0, ERROR_NONE;
+
+	single_read_length: i32;
+	total_read: i64;
+	length := i64(len(data));
+
+	for total_read < length {
+		remaining := length - total_read;
+		MAX :: 1<<32-1;
+		to_read: u32 = min(u32(remaining), MAX);
+
+		e := win32.read_file(win32.Handle(fd), &data[total_read], to_read, &single_read_length, nil);
+		if single_read_length <= 0 || !e {
+			err := Errno(win32.get_last_error());
+			return int(total_read), err;
+		}
+		total_read += i64(single_read_length);
+	}
+	return int(total_read), ERROR_NONE;
+}
+
+seek :: proc(fd: Handle, offset: i64, whence: int) -> (i64, Errno) {
+	w: u32;
+	switch whence {
+	case 0: w = win32.FILE_BEGIN;
+	case 1: w = win32.FILE_CURRENT;
+	case 2: w = win32.FILE_END;
+	}
+	hi := i32(offset>>32);
+	lo := i32(offset);
+	ft := win32.get_file_type(win32.Handle(fd));
+	if ft == win32.FILE_TYPE_PIPE do return 0, ERROR_FILE_IS_PIPE;
+
+	dw_ptr := win32.set_file_pointer(win32.Handle(fd), lo, &hi, w);
+	if dw_ptr == win32.INVALID_SET_FILE_POINTER {
+		err := Errno(win32.get_last_error());
+		return 0, err;
+	}
+	return i64(hi)<<32 + i64(dw_ptr), ERROR_NONE;
+}
+
+file_size :: proc(fd: Handle) -> (i64, Errno) {
+	length: i64;
+	err: Errno;
+	if !win32.get_file_size_ex(win32.Handle(fd), &length) {
+		err = Errno(win32.get_last_error());
+	}
+	return length, err;
+}
+
+
+
+// NOTE(bill): Uses startup to initialize it
+stdin  := get_std_handle(win32.STD_INPUT_HANDLE);
+stdout := get_std_handle(win32.STD_OUTPUT_HANDLE);
+stderr := get_std_handle(win32.STD_ERROR_HANDLE);
+
+
+get_std_handle :: proc(h: int) -> Handle {
+	fd := win32.get_std_handle(i32(h));
+	win32.set_handle_information(fd, win32.HANDLE_FLAG_INHERIT, 0);
+	return Handle(fd);
+}
+
+
+
+
+
+
+last_write_time :: proc(fd: Handle) -> (File_Time, Errno) {
+	file_info: win32.By_Handle_File_Information;
+	if !win32.get_file_information_by_handle(win32.Handle(fd), &file_info) {
+		return 0, Errno(win32.get_last_error());
+	}
+	lo := File_Time(file_info.last_write_time.lo);
+	hi := File_Time(file_info.last_write_time.hi);
+	return lo | hi << 32, ERROR_NONE;
+}
+
+last_write_time_by_name :: proc(name: string) -> (File_Time, Errno) {
+	data: win32.File_Attribute_Data;
+
+	wide_path := win32.utf8_to_wstring(name);
+	if !win32.get_file_attributes_ex_w(wide_path, win32.GetFileExInfoStandard, &data) {
+		return 0, Errno(win32.get_last_error());
+	}
+
+	l := File_Time(data.last_write_time.lo);
+	h := File_Time(data.last_write_time.hi);
+	return l | h << 32, ERROR_NONE;
+}
+
+
+
+heap_alloc :: proc(size: int) -> rawptr {
+	return win32.heap_alloc(win32.get_process_heap(), win32.HEAP_ZERO_MEMORY, size);
+}
+heap_resize :: proc(ptr: rawptr, new_size: int) -> rawptr {
+	if new_size == 0 {
+		heap_free(ptr);
+		return nil;
+	}
+	if ptr == nil do return heap_alloc(new_size);
+
+	return win32.heap_realloc(win32.get_process_heap(), win32.HEAP_ZERO_MEMORY, ptr, new_size);
+}
+heap_free :: proc(ptr: rawptr) {
+	if ptr == nil do return;
+	win32.heap_free(win32.get_process_heap(), 0, ptr);
+}
+
+
+exit :: proc(code: int) -> ! {
+	win32.exit_process(u32(code));
+}
+
+
+
+current_thread_id :: proc "contextless" () -> int {
+	return int(win32.get_current_thread_id());
+}
+
+
+
+_alloc_command_line_arguments :: proc() -> []string {
+	arg_count: i32;
+	arg_list_ptr := win32.command_line_to_argv_w(win32.get_command_line_w(), &arg_count);
+	arg_list := make([]string, int(arg_count));
+	for _, i in arg_list {
+		wc_str := (^win32.Wstring)(uintptr(arg_list_ptr) + size_of(win32.Wstring)*uintptr(i))^;
+		olen := win32.wide_char_to_multi_byte(win32.CP_UTF8, 0, wc_str, -1,
+		                                      nil, 0, nil, nil);
+
+		buf := make([]byte, int(olen));
+		n := win32.wide_char_to_multi_byte(win32.CP_UTF8, 0, wc_str, -1,
+		                                   cstring(&buf[0]), olen, nil, nil);
+		if n > 0 {
+			n -= 1;
+		}
+		arg_list[i] = string(buf[:n]);
+	}
+
+	return arg_list;
+}
+
+get_windows_version_ansi :: proc() -> win32.OS_Version_Info_Ex_A {
+	osvi : win32.OS_Version_Info_Ex_A;
+	osvi.os_version_info_size = size_of(win32.OS_Version_Info_Ex_A);
+    win32.get_version(&osvi);
+    return osvi;
+}
+
+is_windows_xp :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 5 && osvi.minor_version == 1);
+}
+
+is_windows_vista :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 6 && osvi.minor_version == 0);
+}
+
+is_windows_7 :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 6 && osvi.minor_version == 1);
+}
+
+is_windows_8 :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 6 && osvi.minor_version == 2);
+}
+
+is_windows_8_1 :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 6 && osvi.minor_version == 3);
+}
+
+is_windows_10 :: proc() -> bool {
+	osvi := get_windows_version_ansi();
+	return (osvi.major_version == 10 && osvi.minor_version == 0);
+}

core/os_windows.odin

@@ -1,283 +0,0 @@
-#import win32 "sys/windows.odin";
-#import "fmt.odin";
-
-
-Handle    :: int;
-File_Time :: u64;
-Errno     :: int;
-
-INVALID_HANDLE: Handle : -1;
-
-
-O_RDONLY   :: 0x00000;
-O_WRONLY   :: 0x00001;
-O_RDWR     :: 0x00002;
-O_CREAT    :: 0x00040;
-O_EXCL     :: 0x00080;
-O_NOCTTY   :: 0x00100;
-O_TRUNC    :: 0x00200;
-O_NONBLOCK :: 0x00800;
-O_APPEND   :: 0x00400;
-O_SYNC     :: 0x01000;
-O_ASYNC    :: 0x02000;
-O_CLOEXEC  :: 0x80000;
-
-ERROR_NONE:                Errno : 0;
-ERROR_FILE_NOT_FOUND:      Errno : 2;
-ERROR_PATH_NOT_FOUND:      Errno : 3;
-ERROR_ACCESS_DENIED:       Errno : 5;
-ERROR_NO_MORE_FILES:       Errno : 18;
-ERROR_HANDLE_EOF:          Errno : 38;
-ERROR_NETNAME_DELETED:     Errno : 64;
-ERROR_FILE_EXISTS:         Errno : 80;
-ERROR_BROKEN_PIPE:         Errno : 109;
-ERROR_BUFFER_OVERFLOW:     Errno : 111;
-ERROR_INSUFFICIENT_BUFFER: Errno : 122;
-ERROR_MOD_NOT_FOUND:       Errno : 126;
-ERROR_PROC_NOT_FOUND:      Errno : 127;
-ERROR_DIR_NOT_EMPTY:       Errno : 145;
-ERROR_ALREADY_EXISTS:      Errno : 183;
-ERROR_ENVVAR_NOT_FOUND:    Errno : 203;
-ERROR_MORE_DATA:           Errno : 234;
-ERROR_OPERATION_ABORTED:   Errno : 995;
-ERROR_IO_PENDING:          Errno : 997;
-ERROR_NOT_FOUND:           Errno : 1168;
-ERROR_PRIVILEGE_NOT_HELD:  Errno : 1314;
-WSAEACCES:                 Errno : 10013;
-WSAECONNRESET:             Errno : 10054;
-
-// Windows reserves errors >= 1<<29 for application use
-ERROR_FILE_IS_PIPE: Errno : 1<<29 + 0;
-
-
-
-
-open :: proc(path: string, mode: int, perm: u32) -> (Handle, Errno) {
-	using win32;
-	if path.count == 0 {
-		return INVALID_HANDLE, ERROR_FILE_NOT_FOUND;
-	}
-
-	access: u32;
-	match mode & (O_RDONLY|O_WRONLY|O_RDWR) {
-	case O_RDONLY: access = FILE_GENERIC_READ;
-	case O_WRONLY: access = FILE_GENERIC_WRITE;
-	case O_RDWR:   access = FILE_GENERIC_READ | FILE_GENERIC_WRITE;
-	}
-
-	if mode&O_CREAT != 0 {
-		access |= FILE_GENERIC_WRITE;
-	}
-	if mode&O_APPEND != 0 {
-		access &~= FILE_GENERIC_WRITE;
-		access |= FILE_APPEND_DATA;
-	}
-
-	share_mode := cast(u32)(FILE_SHARE_READ|FILE_SHARE_WRITE);
-	sa: ^SECURITY_ATTRIBUTES = nil;
-	sa_inherit := SECURITY_ATTRIBUTES{length = size_of(SECURITY_ATTRIBUTES), inherit_handle = 1};
-	if mode&O_CLOEXEC == 0 {
-		sa = ^sa_inherit;
-	}
-
-	create_mode: u32;
-	match {
-	case mode&(O_CREAT|O_EXCL) == (O_CREAT | O_EXCL):
-		create_mode = CREATE_NEW;
-	case mode&(O_CREAT|O_TRUNC) == (O_CREAT | O_TRUNC):
-		create_mode = CREATE_ALWAYS;
-	case mode&O_CREAT == O_CREAT:
-		create_mode = OPEN_ALWAYS;
-	case mode&O_TRUNC == O_TRUNC:
-		create_mode = TRUNCATE_EXISTING;
-	default:
-		create_mode = OPEN_EXISTING;
-	}
-
-	buf: [300]byte;
-	copy(buf[:], cast([]byte)path);
-
-	handle := cast(Handle)CreateFileA(^buf[0], access, share_mode, sa, create_mode, FILE_ATTRIBUTE_NORMAL, nil);
-	if handle != INVALID_HANDLE {
-		return handle, ERROR_NONE;
-	}
-	err := GetLastError();
-	return INVALID_HANDLE, cast(Errno)err;
-}
-
-close :: proc(fd: Handle) {
-	win32.CloseHandle(cast(win32.HANDLE)fd);
-}
-
-write :: proc(fd: Handle, data: []byte) -> (int, Errno) {
-	bytes_written: i32;
-	e := win32.WriteFile(cast(win32.HANDLE)fd, data.data, cast(i32)data.count, ^bytes_written, nil);
-	if e == win32.FALSE {
-		err := win32.GetLastError();
-		return 0, cast(Errno)err;
-	}
-	return cast(int)bytes_written, ERROR_NONE;
-}
-
-read :: proc(fd: Handle, data: []byte) -> (int, Errno) {
-	bytes_read: i32;
-	e := win32.ReadFile(cast(win32.HANDLE)fd, data.data, cast(u32)data.count, ^bytes_read, nil);
-	if e == win32.FALSE {
-		err := win32.GetLastError();
-		return 0, cast(Errno)err;
-	}
-	return cast(int)bytes_read, ERROR_NONE;
-}
-
-seek :: proc(fd: Handle, offset: i64, whence: int) -> (i64, Errno) {
-	using win32;
-	w: u32;
-	match whence {
-	case 0: w = FILE_BEGIN;
-	case 1: w = FILE_CURRENT;
-	case 2: w = FILE_END;
-	}
-	hi := cast(i32)(offset>>32);
-	lo := cast(i32)(offset);
-	ft := GetFileType(cast(HANDLE)fd);
-	if ft == FILE_TYPE_PIPE {
-		return 0, ERROR_FILE_IS_PIPE;
-	}
-	dw_ptr := SetFilePointer(cast(HANDLE)fd, lo, ^hi, w);
-	if dw_ptr == INVALID_SET_FILE_POINTER {
-		err := GetLastError();
-		return 0, cast(Errno)err;
-	}
-	return cast(i64)hi<<32 + cast(i64)dw_ptr, ERROR_NONE;
-}
-
-
-// NOTE(bill): Uses startup to initialize it
-stdin  := get_std_handle(win32.STD_INPUT_HANDLE);
-stdout := get_std_handle(win32.STD_OUTPUT_HANDLE);
-stderr := get_std_handle(win32.STD_ERROR_HANDLE);
-
-
-get_std_handle :: proc(h: int) -> Handle {
-	fd := win32.GetStdHandle(cast(i32)h);
-	win32.SetHandleInformation(fd, win32.HANDLE_FLAG_INHERIT, 0);
-	return cast(Handle)fd;
-}
-
-
-
-
-
-
-last_write_time :: proc(fd: Handle) -> File_Time {
-	file_info: win32.BY_HANDLE_FILE_INFORMATION;
-	win32.GetFileInformationByHandle(cast(win32.HANDLE)fd, ^file_info);
-	lo := cast(File_Time)file_info.last_write_time.lo;
-	hi := cast(File_Time)file_info.last_write_time.hi;
-	return lo | hi << 32;
-}
-
-last_write_time_by_name :: proc(name: string) -> File_Time {
-	last_write_time: win32.FILETIME;
-	data: win32.FILE_ATTRIBUTE_DATA;
-	buf: [1024]byte;
-
-	assert(buf.count > name.count);
-
-	copy(buf[:], cast([]byte)name);
-
-	if win32.GetFileAttributesExA(^buf[0], win32.GetFileExInfoStandard, ^data) != 0 {
-		last_write_time = data.last_write_time;
-	}
-
-	l := cast(File_Time)last_write_time.lo;
-	h := cast(File_Time)last_write_time.hi;
-	return l | h << 32;
-}
-
-
-
-
-
-read_entire_file :: proc(name: string) -> ([]byte, bool) {
-	buf: [300]byte;
-	copy(buf[:], cast([]byte)name);
-
-	fd, err := open(name, O_RDONLY, 0);
-	if err != ERROR_NONE {
-		return nil, false;
-	}
-	defer close(fd);
-
-	length: i64;
-	file_size_ok := win32.GetFileSizeEx(cast(win32.HANDLE)fd, ^length) != 0;
-	if !file_size_ok {
-		return nil, false;
-	}
-
-	data := new_slice(u8, length);
-	if data.data == nil {
-		return nil, false;
-	}
-
-	single_read_length: i32;
-	total_read: i64;
-
-	for total_read < length {
-		remaining := length - total_read;
-		to_read: u32;
-		MAX :: 1<<32-1;
-		if remaining <= MAX {
-			to_read = cast(u32)remaining;
-		} else {
-			to_read = MAX;
-		}
-
-		win32.ReadFile(cast(win32.HANDLE)fd, ^data[total_read], to_read, ^single_read_length, nil);
-		if single_read_length <= 0 {
-			free(data);
-			return nil, false;
-		}
-
-		total_read += cast(i64)single_read_length;
-	}
-
-	return data, true;
-}
-
-
-
-heap_alloc :: proc(size: int) -> rawptr {
-	assert(size > 0);
-	return win32.HeapAlloc(win32.GetProcessHeap(), win32.HEAP_ZERO_MEMORY, size);
-}
-heap_resize :: proc(ptr: rawptr, new_size: int) -> rawptr {
-	if new_size == 0 {
-		heap_free(ptr);
-		return nil;
-	}
-	if ptr == nil {
-		return heap_alloc(new_size);
-	}
-	return win32.HeapReAlloc(win32.GetProcessHeap(), win32.HEAP_ZERO_MEMORY, ptr, new_size);
-}
-heap_free :: proc(ptr: rawptr) {
-	if ptr == nil {
-		return;
-	}
-	win32.HeapFree(win32.GetProcessHeap(), 0, ptr);
-}
-
-
-exit :: proc(code: int) {
-	win32.ExitProcess(cast(u32)code);
-}
-
-
-
-current_thread_id :: proc() -> int {
-	return cast(int)win32.GetCurrentThreadId();
-}
-
-
-

core/os_x.odin

@@ -1,230 +0,0 @@
-#import "fmt.odin";
-
-Handle    :: i32;
-File_Time :: u64;
-Errno     :: int;
-
-// INVALID_HANDLE: Handle : -1;
-
-
-O_RDONLY   :: 0x00000;
-O_WRONLY   :: 0x00001;
-O_RDWR     :: 0x00002;
-O_CREAT    :: 0x00040;
-O_EXCL     :: 0x00080;
-O_NOCTTY   :: 0x00100;
-O_TRUNC    :: 0x00200;
-O_NONBLOCK :: 0x00800;
-O_APPEND   :: 0x00400;
-O_SYNC     :: 0x01000;
-O_ASYNC    :: 0x02000;
-O_CLOEXEC  :: 0x80000;
-
-// ERROR_NONE:                Errno : 0;
-// ERROR_FILE_NOT_FOUND:      Errno : 2;
-// ERROR_PATH_NOT_FOUND:      Errno : 3;
-// ERROR_ACCESS_DENIED:       Errno : 5;
-// ERROR_NO_MORE_FILES:       Errno : 18;
-// ERROR_HANDLE_EOF:          Errno : 38;
-// ERROR_NETNAME_DELETED:     Errno : 64;
-// ERROR_FILE_EXISTS:         Errno : 80;
-// ERROR_BROKEN_PIPE:         Errno : 109;
-// ERROR_BUFFER_OVERFLOW:     Errno : 111;
-// ERROR_INSUFFICIENT_BUFFER: Errno : 122;
-// ERROR_MOD_NOT_FOUND:       Errno : 126;
-// ERROR_PROC_NOT_FOUND:      Errno : 127;
-// ERROR_DIR_NOT_EMPTY:       Errno : 145;
-// ERROR_ALREADY_EXISTS:      Errno : 183;
-// ERROR_ENVVAR_NOT_FOUND:    Errno : 203;
-// ERROR_MORE_DATA:           Errno : 234;
-// ERROR_OPERATION_ABORTED:   Errno : 995;
-// ERROR_IO_PENDING:          Errno : 997;
-// ERROR_NOT_FOUND:           Errno : 1168;
-// ERROR_PRIVILEGE_NOT_HELD:  Errno : 1314;
-// WSAEACCES:                 Errno : 10013;
-// WSAECONNRESET:             Errno : 10054;
-
-// Windows reserves errors >= 1<<29 for application use
-// ERROR_FILE_IS_PIPE: Errno : 1<<29 + 0;
-
-#foreign_system_library libc "c";
-
-unix_open   :: proc(path: ^u8, mode: int, perm: u32) -> Handle           #foreign libc "open";
-unix_close  :: proc(handle: Handle)                                      #foreign libc "close";
-unix_read   :: proc(handle: Handle, buffer: rawptr, count: int) -> int   #foreign libc "read";
-unix_write  :: proc(handle: Handle, buffer: rawptr, count: int) -> int   #foreign libc "write";
-unix_gettid :: proc() -> u64                                             #foreign libc "gettid";
-
-unix_malloc  :: proc(size: int) -> rawptr                                #foreign libc "malloc";
-unix_free    :: proc(ptr: rawptr)                                        #foreign libc "free";
-unix_realloc :: proc(ptr: rawptr, size: int) -> rawptr                   #foreign libc "realloc";
-
-unix_exit :: proc(status: int)                                           #foreign libc "exit";
-
-
-
-
-open :: proc(path: string, mode: int, perm: u32) -> (Handle, Errno) {
-	return unix_open(path.data, mode, perm), 0;
-}
-
-close :: proc(fd: Handle) {
-	unix_close(fd);
-}
-
-write :: proc(fd: Handle, data: []byte) -> (int, Errno) {
-	return unix_write(fd, data.data, data.count), 0;
-}
-
-read :: proc(fd: Handle, data: []byte) -> (int, Errno) {
-	return unix_read(fd, data.data, data.count), 0;
-}
-
-seek :: proc(fd: Handle, offset: i64, whence: int) -> (i64, Errno) {
-	/*
-	using win32;
-	w: u32;
-	match whence {
-	case 0: w = FILE_BEGIN;
-	case 1: w = FILE_CURRENT;
-	case 2: w = FILE_END;
-	}
-	hi := cast(i32)(offset>>32);
-	lo := cast(i32)(offset);
-	ft := GetFileType(cast(HANDLE)fd);
-	if ft == FILE_TYPE_PIPE {
-		return 0, ERROR_FILE_IS_PIPE;
-	}
-	dw_ptr := SetFilePointer(cast(HANDLE)fd, lo, ^hi, w);
-	if dw_ptr == INVALID_SET_FILE_POINTER {
-		err := GetLastError();
-		return 0, cast(Errno)err;
-	}
-	return cast(i64)hi<<32 + cast(i64)dw_ptr, ERROR_NONE;
-
-	*/
-	return 0, 0;
-}
-
-
-// NOTE(bill): Uses startup to initialize it
-stdin:  Handle = 0; // get_std_handle(win32.STD_INPUT_HANDLE);
-stdout: Handle = 1; // get_std_handle(win32.STD_OUTPUT_HANDLE);
-stderr: Handle = 2; // get_std_handle(win32.STD_ERROR_HANDLE);
-
-
-/*
-get_std_handle :: proc(h: int) -> Handle {
-	fd := win32.GetStdHandle(cast(i32)h);
-	win32.SetHandleInformation(fd, win32.HANDLE_FLAG_INHERIT, 0);
-	return cast(Handle)fd;
-}
-
-
-
-
-
-
-last_write_time :: proc(fd: Handle) -> File_Time {
-	file_info: win32.BY_HANDLE_FILE_INFORMATION;
-	win32.GetFileInformationByHandle(cast(win32.HANDLE)fd, ^file_info);
-	lo := cast(File_Time)file_info.last_write_time.lo;
-	hi := cast(File_Time)file_info.last_write_time.hi;
-	return lo | hi << 32;
-}
-
-last_write_time_by_name :: proc(name: string) -> File_Time {
-	last_write_time: win32.FILETIME;
-	data: win32.FILE_ATTRIBUTE_DATA;
-	buf: [1024]byte;
-
-	assert(buf.count > name.count);
-
-	copy(buf[:], cast([]byte)name);
-
-	if win32.GetFileAttributesExA(^buf[0], win32.GetFileExInfoStandard, ^data) != 0 {
-		last_write_time = data.last_write_time;
-	}
-
-	l := cast(File_Time)last_write_time.lo;
-	h := cast(File_Time)last_write_time.hi;
-	return l | h << 32;
-}
-
-
-
-
-
-read_entire_file :: proc(name: string) -> ([]byte, bool) {
-	buf: [300]byte;
-	copy(buf[:], cast([]byte)name);
-
-	fd, err := open(name, O_RDONLY, 0);
-	if err != ERROR_NONE {
-		return nil, false;
-	}
-	defer close(fd);
-
-	length: i64;
-	file_size_ok := win32.GetFileSizeEx(cast(win32.HANDLE)fd, ^length) != 0;
-	if !file_size_ok {
-		return nil, false;
-	}
-
-	data := new_slice(u8, length);
-	if data.data == nil {
-		return nil, false;
-	}
-
-	single_read_length: i32;
-	total_read: i64;
-
-	for total_read < length {
-		remaining := length - total_read;
-		to_read: u32;
-		MAX :: 1<<32-1;
-		if remaining <= MAX {
-			to_read = cast(u32)remaining;
-		} else {
-			to_read = MAX;
-		}
-
-		win32.ReadFile(cast(win32.HANDLE)fd, ^data[total_read], to_read, ^single_read_length, nil);
-		if single_read_length <= 0 {
-			free(data);
-			return nil, false;
-		}
-
-		total_read += cast(i64)single_read_length;
-	}
-
-	return data, true;
-}
-
-
-*/
-
-heap_alloc :: proc(size: int) -> rawptr {
-	assert(size > 0);
-	return unix_malloc(size);
-}
-heap_resize :: proc(ptr: rawptr, new_size: int) -> rawptr {
-	return unix_realloc(ptr, new_size);
-}
-heap_free :: proc(ptr: rawptr) {
-	unix_free(ptr);
-}
-
-
-exit :: proc(code: int) {
-	unix_exit(code);
-}
-
-
-current_thread_id :: proc() -> int {
-	// return cast(int) unix_gettid();
-	return 0;
-}
-
-
-

core/runtime/internal.odin

@@ -0,0 +1,511 @@
+package runtime
+
+import "core:mem"
+import "core:os"
+import "core:unicode/utf8"
+
+
+print_u64 :: proc(fd: os.Handle, u: u64) {
+	digits := "0123456789";
+
+	a: [129]byte;
+	i := len(a);
+	b := u64(10);
+	for u >= b {
+		i -= 1; a[i] = digits[u % b];
+		u /= b;
+	}
+	i -= 1; a[i] = digits[u % b];
+
+	os.write(fd, a[i:]);
+}
+
+print_i64 :: proc(fd: os.Handle, u: i64) {
+	digits := "0123456789";
+	b :: i64(10);
+
+	neg := u < 0;
+	u = abs(u);
+
+	a: [129]byte;
+	i := len(a);
+	for u >= b {
+		i -= 1; a[i] = digits[u % b];
+		u /= b;
+	}
+	i -= 1; a[i] = digits[u % b];
+	if neg {
+		i -= 1; a[i] = '-';
+	}
+
+	os.write(fd, a[i:]);
+}
+
+print_caller_location :: proc(fd: os.Handle, using loc: Source_Code_Location) {
+	os.write_string(fd, file_path);
+	os.write_byte(fd, '(');
+	print_u64(fd, u64(line));
+	os.write_byte(fd, ':');
+	print_u64(fd, u64(column));
+	os.write_byte(fd, ')');
+}
+print_typeid :: proc(fd: os.Handle, id: typeid) {
+	ti := type_info_of(id);
+	print_type(fd, ti);
+}
+print_type :: proc(fd: os.Handle, ti: ^Type_Info) {
+	if ti == nil {
+		os.write_string(fd, "nil");
+		return;
+	}
+
+	switch info in ti.variant {
+	case Type_Info_Named:
+		os.write_string(fd, info.name);
+	case Type_Info_Integer:
+		switch ti.id {
+		case int:     os.write_string(fd, "int");
+		case uint:    os.write_string(fd, "uint");
+		case uintptr: os.write_string(fd, "uintptr");
+		case:
+			os.write_byte(fd, info.signed ? 'i' : 'u');
+			print_u64(fd, u64(8*ti.size));
+		}
+	case Type_Info_Rune:
+		os.write_string(fd, "rune");
+	case Type_Info_Float:
+		os.write_byte(fd, 'f');
+		print_u64(fd, u64(8*ti.size));
+	case Type_Info_Complex:
+		os.write_string(fd, "complex");
+		print_u64(fd, u64(8*ti.size));
+	case Type_Info_String:
+		os.write_string(fd, "string");
+	case Type_Info_Boolean:
+		switch ti.id {
+		case bool: os.write_string(fd, "bool");
+		case:
+			os.write_byte(fd, 'b');
+			print_u64(fd, u64(8*ti.size));
+		}
+	case Type_Info_Any:
+		os.write_string(fd, "any");
+	case Type_Info_Type_Id:
+		os.write_string(fd, "typeid");
+
+	case Type_Info_Pointer:
+		if info.elem == nil {
+			os.write_string(fd, "rawptr");
+		} else {
+			os.write_string(fd, "^");
+			print_type(fd, info.elem);
+		}
+	case Type_Info_Procedure:
+		os.write_string(fd, "proc");
+		if info.params == nil {
+			os.write_string(fd, "()");
+		} else {
+			t := info.params.variant.(Type_Info_Tuple);
+			os.write_string(fd, "(");
+			for t, i in t.types {
+				if i > 0 do os.write_string(fd, ", ");
+				print_type(fd, t);
+			}
+			os.write_string(fd, ")");
+		}
+		if info.results != nil {
+			os.write_string(fd, " -> ");
+			print_type(fd, info.results);
+		}
+	case Type_Info_Tuple:
+		count := len(info.names);
+		if count != 1 do os.write_string(fd, "(");
+		for name, i in info.names {
+			if i > 0 do os.write_string(fd, ", ");
+
+			t := info.types[i];
+
+			if len(name) > 0 {
+				os.write_string(fd, name);
+				os.write_string(fd, ": ");
+			}
+			print_type(fd, t);
+		}
+		if count != 1 do os.write_string(fd, ")");
+
+	case Type_Info_Array:
+		os.write_string(fd, "[");
+		print_u64(fd, u64(info.count));
+		os.write_string(fd, "]");
+		print_type(fd, info.elem);
+	case Type_Info_Dynamic_Array:
+		os.write_string(fd, "[dynamic]");
+		print_type(fd, info.elem);
+	case Type_Info_Slice:
+		os.write_string(fd, "[]");
+		print_type(fd, info.elem);
+
+	case Type_Info_Map:
+		os.write_string(fd, "map[");
+		print_type(fd, info.key);
+		os.write_byte(fd, ']');
+		print_type(fd, info.value);
+
+	case Type_Info_Struct:
+		os.write_string(fd, "struct ");
+		if info.is_packed    do os.write_string(fd, "#packed ");
+		if info.is_raw_union do os.write_string(fd, "#raw_union ");
+		if info.custom_align {
+			os.write_string(fd, "#align ");
+			print_u64(fd, u64(ti.align));
+			os.write_byte(fd, ' ');
+		}
+		os.write_byte(fd, '{');
+		for name, i in info.names {
+			if i > 0 do os.write_string(fd, ", ");
+			os.write_string(fd, name);
+			os.write_string(fd, ": ");
+			print_type(fd, info.types[i]);
+		}
+		os.write_byte(fd, '}');
+
+	case Type_Info_Union:
+		os.write_string(fd, "union {");
+		for variant, i in info.variants {
+			if i > 0 do os.write_string(fd, ", ");
+			print_type(fd, variant);
+		}
+		os.write_string(fd, "}");
+
+	case Type_Info_Enum:
+		os.write_string(fd, "enum ");
+		print_type(fd, info.base);
+		os.write_string(fd, " {");
+		for name, i in info.names {
+			if i > 0 do os.write_string(fd, ", ");
+			os.write_string(fd, name);
+		}
+		os.write_string(fd, "}");
+
+	case Type_Info_Bit_Field:
+		os.write_string(fd, "bit_field ");
+		if ti.align != 1 {
+			os.write_string(fd, "#align ");
+			print_u64(fd, u64(ti.align));
+			os.write_byte(fd, ' ');
+		}
+		os.write_string(fd, " {");
+		for name, i in info.names {
+			if i > 0 do os.write_string(fd, ", ");
+			os.write_string(fd, name);
+			os.write_string(fd, ": ");
+			print_u64(fd, u64(info.bits[i]));
+		}
+		os.write_string(fd, "}");
+
+	case Type_Info_Bit_Set:
+		os.write_string(fd, "bit_set[");
+
+		switch elem in type_info_base(info.elem).variant {
+		case Type_Info_Enum:
+			print_type(fd, info.elem);
+		case Type_Info_Rune:
+			os.write_encoded_rune(fd, rune(info.lower));
+			os.write_string(fd, "..");
+			os.write_encoded_rune(fd, rune(info.upper));
+		case:
+			print_i64(fd, info.lower);
+			os.write_string(fd, "..");
+			print_i64(fd, info.upper);
+		}
+		if info.underlying != nil {
+			os.write_string(fd, "; ");
+			print_type(fd, info.underlying);
+		}
+		os.write_byte(fd, ']');
+
+	case Type_Info_Opaque:
+		os.write_string(fd, "opaque ");
+		print_type(fd, info.elem);
+
+	case Type_Info_Simd_Vector:
+		if info.is_x86_mmx {
+			os.write_string(fd, "intrinsics.x86_mmx");
+		} else {
+			os.write_string(fd, "intrinsics.vector(");
+			print_u64(fd, u64(info.count));
+			os.write_string(fd, ", ");
+			print_type(fd, info.elem);
+			os.write_byte(fd, ')');
+		}
+	}
+}
+
+string_eq :: proc "contextless" (a, b: string) -> bool {
+	switch {
+	case len(a) != len(b): return false;
+	case len(a) == 0:      return true;
+	case &a[0] == &b[0]:   return true;
+	}
+	return string_cmp(a, b) == 0;
+}
+
+string_cmp :: proc "contextless" (a, b: string) -> int {
+	return mem.compare_byte_ptrs(&a[0], &b[0], min(len(a), len(b)));
+}
+
+string_ne :: inline proc "contextless" (a, b: string) -> bool { return !string_eq(a, b); }
+string_lt :: inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) < 0; }
+string_gt :: inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) > 0; }
+string_le :: inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) <= 0; }
+string_ge :: inline proc "contextless" (a, b: string) -> bool { return string_cmp(a, b) >= 0; }
+
+cstring_len :: proc "contextless" (s: cstring) -> int {
+	n := 0;
+	for p := (^byte)(s); p != nil && p^ != 0; p = mem.ptr_offset(p, 1) {
+		n += 1;
+	}
+	return n;
+}
+
+cstring_to_string :: proc "contextless" (s: cstring) -> string {
+	if s == nil do return "";
+	ptr := (^byte)(s);
+	n := cstring_len(s);
+	return transmute(string)mem.Raw_String{ptr, n};
+}
+
+
+complex64_eq :: inline proc "contextless"  (a, b: complex64)  -> bool { return real(a) == real(b) && imag(a) == imag(b); }
+complex64_ne :: inline proc "contextless"  (a, b: complex64)  -> bool { return real(a) != real(b) || imag(a) != imag(b); }
+
+complex128_eq :: inline proc "contextless" (a, b: complex128) -> bool { return real(a) == real(b) && imag(a) == imag(b); }
+complex128_ne :: inline proc "contextless" (a, b: complex128) -> bool { return real(a) != real(b) || imag(a) != imag(b); }
+
+
+
+
+bounds_check_error :: proc "contextless" (file: string, line, column: int, index, count: int) {
+	if 0 <= index && index < count do return;
+	handle_error :: proc "contextless" (file: string, line, column: int, index, count: int) {
+		fd := os.stderr;
+		print_caller_location(fd, Source_Code_Location{file, line, column, "", 0});
+		os.write_string(fd, " Index ");
+		print_i64(fd, i64(index));
+		os.write_string(fd, " is out of bounds range 0:");
+		print_i64(fd, i64(count));
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(file, line, column, index, count);
+}
+
+slice_handle_error :: proc "contextless" (file: string, line, column: int, lo, hi: int, len: int) {
+	fd := os.stderr;
+	print_caller_location(fd, Source_Code_Location{file, line, column, "", 0});
+	os.write_string(fd, " Invalid slice indices: ");
+	print_i64(fd, i64(lo));
+	os.write_string(fd, ":");
+	print_i64(fd, i64(hi));
+	os.write_string(fd, ":");
+	print_i64(fd, i64(len));
+	os.write_byte(fd, '\n');
+	debug_trap();
+}
+
+slice_expr_error_hi :: proc "contextless" (file: string, line, column: int, hi: int, len: int) {
+	if 0 <= hi && hi <= len do return;
+	slice_handle_error(file, line, column, 0, hi, len);
+}
+
+slice_expr_error_lo_hi :: proc "contextless" (file: string, line, column: int, lo, hi: int, len: int) {
+	if 0 <= lo && lo <= len && lo <= hi && hi <= len do return;
+	slice_handle_error(file, line, column, lo, hi, len);
+}
+
+dynamic_array_expr_error :: proc "contextless" (file: string, line, column: int, low, high, max: int) {
+	if 0 <= low && low <= high && high <= max do return;
+	handle_error :: proc "contextless" (file: string, line, column: int, low, high, max: int) {
+		fd := os.stderr;
+		print_caller_location(fd, Source_Code_Location{file, line, column, "", 0});
+		os.write_string(fd, " Invalid dynamic array values: ");
+		print_i64(fd, i64(low));
+		os.write_string(fd, ":");
+		print_i64(fd, i64(high));
+		os.write_string(fd, ":");
+		print_i64(fd, i64(max));
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(file, line, column, low, high, max);
+}
+
+
+type_assertion_check :: proc "contextless" (ok: bool, file: string, line, column: int, from, to: typeid) {
+	if ok do return;
+	handle_error :: proc "contextless" (file: string, line, column: int, from, to: typeid) {
+		fd := os.stderr;
+		print_caller_location(fd, Source_Code_Location{file, line, column, "", 0});
+		os.write_string(fd, " Invalid type assertion from ");
+		print_typeid(fd, from);
+		os.write_string(fd, " to ");
+		print_typeid(fd, to);
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(file, line, column, from, to);
+}
+
+string_decode_rune :: inline proc "contextless" (s: string) -> (rune, int) {
+	return utf8.decode_rune_in_string(s);
+}
+
+bounds_check_error_loc :: inline proc "contextless" (using loc := #caller_location, index, count: int) {
+	bounds_check_error(file_path, int(line), int(column), index, count);
+}
+
+slice_expr_error_hi_loc :: inline proc "contextless" (using loc := #caller_location, hi: int, len: int) {
+	slice_expr_error_hi(file_path, int(line), int(column), hi, len);
+}
+
+slice_expr_error_lo_hi_loc :: inline proc "contextless" (using loc := #caller_location, lo, hi: int, len: int) {
+	slice_expr_error_lo_hi(file_path, int(line), int(column), lo, hi, len);
+}
+
+dynamic_array_expr_error_loc :: inline proc "contextless" (using loc := #caller_location, low, high, max: int) {
+	dynamic_array_expr_error(file_path, int(line), int(column), low, high, max);
+}
+
+
+make_slice_error_loc :: inline proc "contextless" (loc := #caller_location, len: int) {
+	if 0 <= len do return;
+	handle_error :: proc "contextless" (loc: Source_Code_Location, len: int) {
+		fd := os.stderr;
+		print_caller_location(fd, loc);
+		os.write_string(fd, " Invalid slice length for make: ");
+		print_i64(fd, i64(len));
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(loc, len);
+}
+
+make_dynamic_array_error_loc :: inline proc "contextless" (using loc := #caller_location, len, cap: int) {
+	if 0 <= len && len <= cap do return;
+	handle_error :: proc "contextless" (loc: Source_Code_Location, len, cap: int) {
+		fd := os.stderr;
+		print_caller_location(fd, loc);
+		os.write_string(fd, " Invalid dynamic array parameters for make: ");
+		print_i64(fd, i64(len));
+		os.write_byte(fd, ':');
+		print_i64(fd, i64(cap));
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(loc, len, cap);
+}
+
+make_map_expr_error_loc :: inline proc "contextless" (loc := #caller_location, cap: int) {
+	if 0 <= cap do return;
+	handle_error :: proc "contextless" (loc: Source_Code_Location, cap: int) {
+		fd := os.stderr;
+		print_caller_location(fd, loc);
+		os.write_string(fd, " Invalid map capacity for make: ");
+		print_i64(fd, i64(cap));
+		os.write_byte(fd, '\n');
+		debug_trap();
+	}
+	handle_error(loc, cap);
+}
+
+
+
+
+@(default_calling_convention = "c")
+foreign {
+	@(link_name="llvm.sqrt.f32") _sqrt_f32 :: proc(x: f32) -> f32 ---
+	@(link_name="llvm.sqrt.f64") _sqrt_f64 :: proc(x: f64) -> f64 ---
+}
+abs_f32 :: inline proc "contextless" (x: f32) -> f32 {
+	foreign {
+		@(link_name="llvm.fabs.f32") _abs :: proc "c" (x: f32) -> f32 ---
+	}
+	return _abs(x);
+}
+abs_f64 :: inline proc "contextless" (x: f64) -> f64 {
+	foreign {
+		@(link_name="llvm.fabs.f64") _abs :: proc "c" (x: f64) -> f64 ---
+	}
+	return _abs(x);
+}
+
+min_f32 :: proc(a, b: f32) -> f32 {
+	foreign {
+		@(link_name="llvm.minnum.f32") _min :: proc "c" (a, b: f32) -> f32 ---
+	}
+	return _min(a, b);
+}
+min_f64 :: proc(a, b: f64) -> f64 {
+	foreign {
+		@(link_name="llvm.minnum.f64") _min :: proc "c" (a, b: f64) -> f64 ---
+	}
+	return _min(a, b);
+}
+max_f32 :: proc(a, b: f32) -> f32 {
+	foreign {
+		@(link_name="llvm.maxnum.f32") _max :: proc "c" (a, b: f32) -> f32 ---
+	}
+	return _max(a, b);
+}
+max_f64 :: proc(a, b: f64) -> f64 {
+	foreign {
+		@(link_name="llvm.maxnum.f64") _max :: proc "c" (a, b: f64) -> f64 ---
+	}
+	return _max(a, b);
+}
+
+abs_complex64 :: inline proc "contextless" (x: complex64) -> f32 {
+	r, i := real(x), imag(x);
+	return _sqrt_f32(r*r + i*i);
+}
+abs_complex128 :: inline proc "contextless" (x: complex128) -> f64 {
+	r, i := real(x), imag(x);
+	return _sqrt_f64(r*r + i*i);
+}
+
+
+quo_complex64 :: proc(n, m: complex64) -> complex64 {
+	e, f: f32;
+
+	if abs(real(m)) >= abs(imag(m)) {
+		ratio := imag(m) / real(m);
+		denom := real(m) + ratio*imag(m);
+		e = (real(n) + imag(n)*ratio) / denom;
+		f = (imag(n) - real(n)*ratio) / denom;
+	} else {
+		ratio := real(m) / imag(m);
+		denom := imag(m) + ratio*real(m);
+		e = (real(n)*ratio + imag(n)) / denom;
+		f = (imag(n)*ratio - real(n)) / denom;
+	}
+
+	return complex(e, f);
+}
+
+quo_complex128 :: proc(n, m: complex128) -> complex128 {
+	e, f: f64;
+
+	if abs(real(m)) >= abs(imag(m)) {
+		ratio := imag(m) / real(m);
+		denom := real(m) + ratio*imag(m);
+		e = (real(n) + imag(n)*ratio) / denom;
+		f = (imag(n) - real(n)*ratio) / denom;
+	} else {
+		ratio := real(m) / imag(m);
+		denom := imag(m) + ratio*real(m);
+		e = (real(n)*ratio + imag(n)) / denom;
+		f = (imag(n)*ratio - real(n)) / denom;
+	}
+
+	return complex(e, f);
+}

core/runtime/internal_windows.odin

@@ -0,0 +1,202 @@
+package runtime
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.cttz.i8")  _ctz_u8  :: proc(i:  u8,  is_zero_undef := false) ->  u8 ---
+	@(link_name="llvm.cttz.i16") _ctz_u16 :: proc(i: u16,  is_zero_undef := false) -> u16 ---
+	@(link_name="llvm.cttz.i32") _ctz_u32 :: proc(i: u32,  is_zero_undef := false) -> u32 ---
+	@(link_name="llvm.cttz.i64") _ctz_u64 :: proc(i: u64,  is_zero_undef := false) -> u64 ---
+}
+_ctz :: proc{
+	_ctz_u8,
+	_ctz_u16,
+	_ctz_u32,
+	_ctz_u64,
+};
+
+@(default_calling_convention="none")
+foreign {
+	@(link_name="llvm.ctlz.i8")  _clz_u8  :: proc(i:  u8,  is_zero_undef := false) ->  u8 ---
+	@(link_name="llvm.ctlz.i16") _clz_u16 :: proc(i: u16,  is_zero_undef := false) -> u16 ---
+	@(link_name="llvm.ctlz.i32") _clz_u32 :: proc(i: u32,  is_zero_undef := false) -> u32 ---
+	@(link_name="llvm.ctlz.i64") _clz_u64 :: proc(i: u64,  is_zero_undef := false) -> u64 ---
+}
+_clz :: proc{
+	_clz_u8,
+	_clz_u16,
+	_clz_u32,
+	_clz_u64,
+};
+
+
+udivmod128 :: proc "c" (a, b: u128, rem: ^u128) -> u128 {
+	n := transmute([2]u64)a;
+	d := transmute([2]u64)b;
+	q, r: [2]u64 = ---, ---;
+	sr: u32 = 0;
+
+	low  :: ODIN_ENDIAN == "big" ? 1 : 0;
+	high :: 1 - low;
+	U64_BITS :: 8*size_of(u64);
+	U128_BITS :: 8*size_of(u128);
+
+	// Special Cases
+
+	if n[high] == 0 {
+		if d[high] == 0 {
+			if rem != nil {
+				rem^ = u128(n[low] % d[low]);
+			}
+			return u128(n[low] / d[low]);
+		}
+
+		if rem != nil {
+			rem^ = u128(n[low]);
+		}
+		return 0;
+	}
+
+	if d[low] == 0 {
+		if d[high] == 0 {
+			if rem != nil {
+				rem^ = u128(n[high] % d[low]);
+			}
+			return u128(n[high] / d[low]);
+		}
+		if n[low] == 0 {
+			if rem != nil {
+				r[high] = n[high] % d[high];
+				r[low] = 0;
+				rem^ = transmute(u128)r;
+			}
+			return u128(n[high] / d[high]);
+		}
+
+		if d[high] & (d[high]-1) == 0 {
+			if rem != nil {
+				r[low] = n[low];
+				r[high] = n[high] & (d[high] - 1);
+				rem^ = transmute(u128)r;
+			}
+			return u128(n[high] >> _ctz(d[high]));
+		}
+
+		sr = transmute(u32)(i32(_clz(d[high])) - i32(_clz(n[high])));
+		if sr > U64_BITS - 2 {
+			if rem != nil {
+				rem^ = a;
+			}
+			return 0;
+		}
+
+		sr += 1;
+
+		q[low]  = 0;
+		q[high] = n[low] << u64(U64_BITS - sr);
+		r[high] = n[high] >> sr;
+		r[low]  = (n[high] << (U64_BITS - sr)) | (n[low] >> sr);
+	} else {
+		if d[high] == 0 {
+			if d[low] & (d[low] - 1) == 0 {
+				if rem != nil {
+					rem^ = u128(n[low] & (d[low] - 1));
+				}
+				if d[low] == 1 {
+					return a;
+				}
+				sr = u32(_ctz(d[low]));
+				q[high] = n[high] >> sr;
+				q[low] = (n[high] << (U64_BITS-sr)) | (n[low] >> sr);
+				return transmute(u128)q;
+			}
+
+			sr = 1 + U64_BITS + u32(_clz(d[low])) - u32(_clz(n[high]));
+
+			switch {
+			case sr == U64_BITS:
+				q[low]  = 0;
+				q[high] = n[low];
+				r[high] = 0;
+				r[low]  = n[high];
+			case sr < U64_BITS:
+				q[low]  = 0;
+				q[high] = n[low] << (U64_BITS - sr);
+				r[high] = n[high] >> sr;
+				r[low]  = (n[high] << (U64_BITS - sr)) | (n[low] >> sr);
+			case:
+				q[low]  = n[low] << (U128_BITS - sr);
+				q[high] = (n[high] << (U128_BITS - sr)) | (n[low] >> (sr - U64_BITS));
+				r[high] = 0;
+				r[low]  = n[high] >> (sr - U64_BITS);
+			}
+		} else {
+			sr = transmute(u32)(i32(_clz(d[high])) - i32(_clz(n[high])));
+
+			if sr > U64_BITS - 1 {
+				if rem != nil {
+					rem^ = a;
+				}
+				return 0;
+			}
+
+			sr += 1;
+
+			q[low] = 0;
+			if sr == U64_BITS {
+				q[high] = n[low];
+				r[high] = 0;
+				r[low]  = n[high];
+			} else {
+				r[high] = n[high] >> sr;
+				r[low]  = (n[high] << (U64_BITS - sr)) | (n[low] >> sr);
+				q[high] = n[low] << (U64_BITS - sr);
+			}
+		}
+	}
+
+	carry: u32 = 0;
+	r_all: u128 = ---;
+
+	for ; sr > 0; sr -= 1 {
+		r[high] = (r[high] << 1) | (r[low]  >> (U64_BITS - 1));
+		r[low]  = (r[low]  << 1) | (q[high] >> (U64_BITS - 1));
+		q[high] = (q[high] << 1) | (q[low]  >> (U64_BITS - 1));
+		q[low]  = (q[low]  << 1) | u64(carry);
+
+		r_all = transmute(u128)r;
+		s := i128(b - r_all - 1) >> (U128_BITS - 1);
+		carry = u32(s & 1);
+		r_all -= b & transmute(u128)s;
+		r = transmute([2]u64)r_all;
+	}
+
+	q_all := ((transmute(u128)q) << 1) | u128(carry);
+	if rem != nil {
+		rem^ = r_all;
+	}
+
+	return q_all;
+}
+
+@(link_name="__umodti3")
+umodti3 :: proc "c" (a, b: i128) -> i128 {
+	s_a := a >> (128 - 1);
+	s_b := b >> (128 - 1);
+	an := (a ~ s_a) - s_a;
+	bn := (b ~ s_b) - s_b;
+
+	r: u128 = ---;
+	_ = udivmod128(transmute(u128)an, transmute(u128)bn, &r);
+	return (transmute(i128)r ~ s_a) - s_a;
+}
+
+
+@(link_name="__udivmodti4")
+udivmodti4 :: proc "c" (a, b: u128, rem: ^u128) -> u128 {
+	return udivmod128(a, b, rem);
+}
+
+@(link_name="__udivti3")
+udivti3 :: proc "c" (a, b: u128) -> u128 {
+	return udivmodti4(a, b, nil);
+}

core/runtime/procs_windows_amd64.odin

@@ -0,0 +1,50 @@
+package runtime
+
+foreign import kernel32 "system:Kernel32.lib"
+
+@(link_name="memcpy")
+memcpy :: proc "c" (dst, src: rawptr, len: int) -> rawptr {
+	foreign kernel32 {
+		RtlCopyMemory :: proc "c" (dst, src: rawptr, len: int) ---
+	}
+	RtlCopyMemory(dst, src, len);
+	return dst;
+}
+
+@(link_name="memmove")
+memmove :: proc "c" (dst, src: rawptr, len: int) -> rawptr {
+	foreign kernel32 {
+		RtlMoveMemory :: proc "c" (dst, src: rawptr, len: int) ---
+	}
+	RtlMoveMemory(dst, src, len);
+	return dst;
+}
+
+@(link_name="memset")
+memset :: proc "c" (ptr: rawptr, val: i32, len: int) -> rawptr {
+	foreign kernel32 {
+		RtlFillMemory :: proc "c" (dst: rawptr, len: int, fill: byte) ---
+	}
+	RtlFillMemory(ptr, len, byte(val));
+	return ptr;
+}
+
+// @(link_name="memcmp")
+// memcmp :: proc "c" (dst, src: rawptr, len: int) -> i32 {
+// 	if dst == nil || src == nil {
+// 		return 0;
+// 	}
+// 	if dst == src {
+// 		return 0;
+// 	}
+// 	d, s := uintptr(dst), uintptr(src);
+// 	n := uintptr(len);
+
+// 	for i := uintptr(0); i < n; i += 1 {
+// 		x, y := (^byte)(d+i)^, (^byte)(s+i)^;
+// 		if x != y {
+// 			return x < y ? -1 : +1;
+// 		}
+// 	}
+// 	return 0;
+// }

core/sort/sort.odin

@@ -0,0 +1,225 @@
+package sort
+
+import "core:mem"
+
+bubble_sort_proc :: proc(array: $A/[]$T, f: proc(T, T) -> int) {
+	assert(f != nil);
+	count := len(array);
+
+	init_j, last_j := 0, count-1;
+
+	for {
+		init_swap, prev_swap := -1, -1;
+
+		for j in init_j..<last_j {
+			if f(array[j], array[j+1]) > 0 {
+				array[j], array[j+1] = array[j+1], array[j];
+				prev_swap = j;
+				if init_swap == -1 do init_swap = j;
+			}
+		}
+
+		if prev_swap == -1 do return;
+
+		init_j = max(init_swap-1, 0);
+		last_j = prev_swap;
+	}
+}
+
+bubble_sort :: proc(array: $A/[]$T) {
+	count := len(array);
+
+	init_j, last_j := 0, count-1;
+
+	for {
+		init_swap, prev_swap := -1, -1;
+
+		for j in init_j..<last_j {
+			if array[j] > array[j+1] {
+				array[j], array[j+1] = array[j+1], array[j];
+				prev_swap = j;
+				if init_swap == -1 do init_swap = j;
+			}
+		}
+
+		if prev_swap == -1 do return;
+
+		init_j = max(init_swap-1, 0);
+		last_j = prev_swap;
+	}
+}
+
+quick_sort_proc :: proc(array: $A/[]$T, f: proc(T, T) -> int) {
+	assert(f != nil);
+	a := array;
+	n := len(a);
+	if n < 2 do return;
+
+	p := a[n/2];
+	i, j := 0, n-1;
+
+	loop: for {
+		for f(a[i], p) < 0 do i += 1;
+		for f(p, a[j]) < 0 do j -= 1;
+
+		if i >= j do break loop;
+
+		a[i], a[j] = a[j], a[i];
+		i += 1;
+		j -= 1;
+	}
+
+	quick_sort_proc(a[0:i], f);
+	quick_sort_proc(a[i:n], f);
+}
+
+quick_sort :: proc(array: $A/[]$T) {
+	a := array;
+	n := len(a);
+	if n < 2 do return;
+
+	p := a[n/2];
+	i, j := 0, n-1;
+
+	loop: for {
+		for a[i] < p do i += 1;
+		for p < a[j] do j -= 1;
+
+		if i >= j do break loop;
+
+		a[i], a[j] = a[j], a[i];
+		i += 1;
+		j -= 1;
+	}
+
+	quick_sort(a[0:i]);
+	quick_sort(a[i:n]);
+}
+
+_log2 :: proc(n: int) -> int {
+	res := 0;
+	for ; n != 0; n >>= 1 do res += 1;
+	return res;
+}
+
+merge_sort_proc :: proc(array: $A/[]$T, f: proc(T, T) -> int) {
+	merge_slices :: proc(arr1, arr2, out: A, f: proc(T, T) -> int) {
+		N1, N2 := len(arr1), len(arr2);
+		i, j := 0, 0;
+		for k in 0..<N1+N2 {
+			if j == N2 || i < N1 && j < N2 && f(arr1[i], arr2[j]) < 0 {
+				out[k] = arr1[i];
+				i += 1;
+			} else {
+				out[k] = arr2[j];
+				j += 1;
+			}
+		}
+	}
+
+	assert(f != nil);
+
+	arr1 := array;
+	N := len(arr1);
+	arr2 := make([]T, N);
+	defer free(arr2);
+
+	a, b, m, M := N/2, N, 1, _log2(N);
+
+	for i in 0..M {
+		for j in 0..<a {
+			k := 2*j*m;
+			merge_slices(arr1[k:k+m], arr1[k+m:k+m+m], arr2[k:], f);
+		}
+		if N-b > m {
+			k := 2*a*m;
+			merge_slices(arr1[k:k+m], arr1[k+m : k+m+(N-b)&(m-1)], arr2[k:], f);
+		} else {
+			copy(arr2[b:N], arr1[b:N]);
+		}
+		arr1, arr2 = arr2, arr1;
+		m <<= 1;
+		a >>= 1;
+		b = a << uint(i) << 2;
+	}
+
+	if M & 1 == 0 do copy(arr2, arr1);
+}
+
+merge_sort :: proc(array: $A/[]$T) {
+	merge_slices :: proc(arr1, arr2, out: A) {
+		N1, N2 := len(arr1), len(arr2);
+		i, j := 0, 0;
+		for k in 0..<N1+N2 {
+			if j == N2 || i < N1 && j < N2 && arr1[i] < arr2[j] {
+				out[k] = arr1[i];
+				i += 1;
+			} else {
+				out[k] = arr2[j];
+				j += 1;
+			}
+		}
+	}
+
+	arr1 := array;
+	N := len(arr1);
+	arr2 := make([]T, N);
+	defer free(arr2);
+
+	a, b, m, M := N/2, N, 1, _log2(N);
+
+	for i in 0..M {
+		for j in 0..<a {
+			k := 2*j*m;
+			merge_slices(arr1[k:k+m], arr1[k+m:k+m+m], arr2[k:]);
+		}
+		if N-b > m {
+			k := 2*a*m;
+			merge_slices(arr1[k:k+m], arr1[k+m : k+m+(N-b)&(m-1)], arr2[k:]);
+		} else {
+			copy(arr2[b:N], arr1[b:N]);
+		}
+		arr1, arr2 = arr2, arr1;
+		m <<= 1;
+		a >>= 1;
+		b = a << uint(i) << 2;
+	}
+
+	if M & 1 == 0 do copy(arr2, arr1);
+}
+
+
+compare_bools :: proc(a, b: bool) -> int {
+	switch {
+	case !a && b: return -1;
+	case a && !b: return +1;
+	}
+	return 0;
+}
+
+
+compare_ints :: proc(a, b: int) -> int {
+	switch delta := a - b; {
+	case delta < 0: return -1;
+	case delta > 0: return +1;
+	}
+	return 0;
+}
+
+compare_f32s :: proc(a, b: f32) -> int {
+	switch delta := a - b; {
+	case delta < 0: return -1;
+	case delta > 0: return +1;
+	}
+	return 0;
+}
+compare_f64s :: proc(a, b: f64) -> int {
+	switch delta := a - b; {
+	case delta < 0: return -1;
+	case delta > 0: return +1;
+	}
+	return 0;
+}
+compare_strings :: proc(a, b: string) -> int {
+	return mem.compare_byte_ptrs(&a[0], &b[0], min(len(a), len(b)));
+}

core/strconv/generic_float.odin

@@ -0,0 +1,437 @@
+package strconv
+
+using import "core:decimal"
+
+Int_Flag :: enum {
+	Prefix,
+	Plus,
+	Space,
+}
+Int_Flags :: bit_set[Int_Flag];
+
+Decimal_Slice :: struct {
+	digits:        []byte,
+	count:         int,
+	decimal_point: int,
+	neg:           bool,
+}
+
+Float_Info :: struct {
+	mantbits: uint,
+	expbits:  uint,
+	bias:     int,
+}
+
+
+_f16_info := Float_Info{10, 5,   -15};
+_f32_info := Float_Info{23, 8,  -127};
+_f64_info := Float_Info{52, 11, -1023};
+
+
+generic_ftoa :: proc(buf: []byte, val: f64, fmt: byte, prec, bit_size: int) -> []byte {
+	bits: u64;
+	flt: ^Float_Info;
+	switch bit_size {
+	case 32:
+		bits = u64(transmute(u32)f32(val));
+		flt = &_f32_info;
+	case 64:
+		bits = transmute(u64)val;
+		flt = &_f64_info;
+	case:
+		panic("strconv: invalid bit_size");
+	}
+
+	neg  := bits>>(flt.expbits+flt.mantbits) != 0;
+	exp  := int(bits>>flt.mantbits) & (1<<flt.expbits - 1);
+	mant := bits & (u64(1) << flt.mantbits - 1);
+
+	switch exp {
+	case 1<<flt.expbits - 1:
+		s: string;
+		if mant != 0 {
+			s = "NaN";
+		} else if neg {
+			s = "-Inf";
+		} else {
+			s = "+Inf";
+		}
+		n := copy(buf, cast([]byte)s);
+		return buf[:n];
+
+	case 0: // denormalized
+		exp += 1;
+
+	case:
+		mant |= u64(1) << flt.mantbits;
+	}
+
+	exp += flt.bias;
+
+	d_: Decimal;
+	d := &d_;
+	assign(d, mant);
+	shift(d, exp - int(flt.mantbits));
+	digs: Decimal_Slice;
+	shortest := prec < 0;
+	if shortest {
+		round_shortest(d, mant, exp, flt);
+		digs = Decimal_Slice{digits = d.digits[:], count = d.count, decimal_point = d.decimal_point};
+		switch fmt {
+		case 'e', 'E': prec = digs.count-1;
+		case 'f', 'F': prec = max(digs.count-digs.decimal_point, 0);
+		case 'g', 'G': prec = digs.count;
+		}
+	} else {
+		switch fmt {
+		case 'e', 'E': round(d, prec+1);
+		case 'f', 'F': round(d, d.decimal_point+prec);
+		case 'g', 'G':
+			if prec == 0 {
+				prec = 1;
+			}
+			round(d, prec);
+		}
+
+		digs = Decimal_Slice{digits = d.digits[:], count = d.count, decimal_point = d.decimal_point};
+	}
+	return format_digits(buf, shortest, neg, digs, prec, fmt);
+}
+
+
+
+format_digits :: proc(buf: []byte, shortest: bool, neg: bool, digs: Decimal_Slice, prec: int, fmt: byte) -> []byte {
+	Buffer :: struct {
+		b: []byte,
+		n: int,
+	}
+
+	to_bytes :: proc(b: Buffer) -> []byte do return b.b[:b.n];
+	add_bytes :: proc(buf: ^Buffer, bytes: ..byte) {
+		buf.n += copy(buf.b[buf.n:], bytes);
+	}
+
+	b := Buffer{b = buf};
+
+	switch fmt {
+	case 'f', 'F':
+		add_bytes(&b, neg ? '-' : '+');
+
+		// integer, padded with zeros when needed
+		if digs.decimal_point > 0 {
+			m := min(digs.count, digs.decimal_point);
+			add_bytes(&b, ..digs.digits[0:m]);
+			for ; m < digs.decimal_point; m += 1 {
+				add_bytes(&b, '0');
+			}
+		} else {
+			add_bytes(&b, '0');
+		}
+
+
+		// fractional part
+		if prec > 0 {
+			add_bytes(&b, '.');
+			for i in 0..<prec {
+				c: byte = '0';
+				if j := digs.decimal_point + i; 0 <= j && j < digs.count {
+					c = digs.digits[j];
+				}
+				add_bytes(&b, c);
+			}
+		}
+		return to_bytes(b);
+
+	case 'e', 'E':
+		add_bytes(&b, neg ? '-' : '+');
+
+		ch := byte('0');
+		if digs.count != 0 {
+			ch = digs.digits[0];
+		}
+		add_bytes(&b, ch);
+
+		if prec > 0 {
+			add_bytes(&b, '.');
+			i := 1;
+			m := min(digs.count, prec+1);
+			if i < m {
+				add_bytes(&b, ..digs.digits[i:m]);
+				i = m;
+			}
+			for ; i <= prec; i += 1 {
+				add_bytes(&b, '0');
+			}
+		}
+
+		add_bytes(&b, fmt);
+		exp := digs.decimal_point-1;
+		if digs.count == 0 {
+			// Zero has exponent of 0
+			exp = 0;
+		}
+
+		ch = '+';
+		if exp < 0 {
+			ch = '-';
+			exp = -exp;
+		}
+		add_bytes(&b, ch);
+
+		switch {
+		case exp < 10:  add_bytes(&b, '0', byte(exp)+'0'); // add prefix 0
+		case exp < 100: add_bytes(&b, byte(exp/10)+'0',  byte(exp%10)+'0');
+		case:           add_bytes(&b, byte(exp/100)+'0', byte(exp/10)%10+'0', byte(exp%10)+'0');
+		}
+
+		return to_bytes(b);
+
+	case 'g', 'G':
+		eprec := prec;
+		if eprec > digs.count && digs.count >= digs.decimal_point {
+			eprec = digs.count;
+		}
+
+		if shortest {
+			eprec = 6;
+		}
+
+		exp := digs.decimal_point - 1;
+		if exp < -4 || exp >= eprec {
+			if prec > digs.count {
+				prec = digs.count;
+			}
+			return format_digits(buf, shortest, neg, digs, prec-1, fmt+'e'-'g'); // keep the same case
+		}
+
+		if prec > digs.decimal_point {
+			prec = digs.count;
+		}
+
+		return format_digits(buf, shortest, neg, digs, max(prec-digs.decimal_point, 0), 'f');
+
+	case:
+		add_bytes(&b, '%', fmt);
+		return to_bytes(b);
+	}
+
+
+}
+
+round_shortest :: proc(d: ^Decimal, mant: u64, exp: int, flt: ^Float_Info) {
+	if mant == 0 { // If mantissa is zero, the number is zero
+		d.count = 0;
+		return;
+	}
+
+	/*
+		10^(dp-nd) > 2^(exp-mantbits)
+		log2(10) * (dp-nd) > exp-mantbits
+		log(2) >~ 0.332
+		332*(dp-nd) >= 100*(exp-mantbits)
+	 */
+	minexp := flt.bias+1;
+	if exp > minexp && 332*(d.decimal_point-d.count) >= 100*(exp - int(flt.mantbits)) {
+		// Number is already its shortest
+		return;
+	}
+
+	upper_: Decimal; upper := &upper_;
+	assign(upper, 2*mant - 1);
+	shift(upper, exp - int(flt.mantbits) - 1);
+
+	mantlo: u64;
+	explo:  int;
+	if mant > 1<<flt.mantbits || exp == minexp {
+		mantlo = mant-1;
+		explo = exp;
+	} else {
+		mantlo = 2*mant - 1;
+		explo = exp-1;
+	}
+	lower_: Decimal; lower := &lower_;
+	assign(lower, 2*mantlo + 1);
+	shift(lower, explo - int(flt.mantbits) - 1);
+
+	inclusive := mant%2 == 0;
+
+	for i in 0..<d.count {
+		l: byte = '0'; // lower digit
+		if i < lower.count {
+			l = lower.digits[i];
+		}
+		m := d.digits[i];   // middle digit
+		u: byte = '0'; // upper digit
+		if i < upper.count {
+			u = upper.digits[i];
+		}
+
+		ok_round_down := l != m || inclusive && i+1 == lower.count;
+		ok_round_up   := m != u && (inclusive || m+1 < u || i+1 < upper.count);
+
+		if ok_round_down && ok_round_up {
+			round(d, i+1);
+			return;
+		}
+		if ok_round_down {
+			round_down(d, i+1);
+			return;
+		}
+		if ok_round_up {
+			round_up(d, i+1);
+			return;
+		}
+	}
+
+}
+
+MAX_BASE :: 32;
+digits := "0123456789abcdefghijklmnopqrstuvwxyz";
+
+
+is_integer_negative :: proc(u: u64, is_signed: bool, bit_size: int) -> (unsigned: u64, neg: bool) {
+	if is_signed {
+		switch bit_size {
+		case 8:
+			i := i8(u);
+			neg = i < 0;
+			u = u64(abs(i64(i)));
+		case 16:
+			i := i16(u);
+			neg = i < 0;
+			u = u64(abs(i64(i)));
+		case 32:
+			i := i32(u);
+			neg = i < 0;
+			u = u64(abs(i64(i)));
+		case 64:
+			i := i64(u);
+			neg = i < 0;
+			u = u64(abs(i64(i)));
+		case:
+			panic("is_integer_negative: Unknown integer size");
+		}
+	}
+	return u, neg;
+}
+
+append_bits :: proc(buf: []byte, u: u64, base: int, is_signed: bool, bit_size: int, digits: string, flags: Int_Flags) -> string {
+	if base < 2 || base > MAX_BASE {
+		panic("strconv: illegal base passed to append_bits");
+	}
+
+	neg: bool;
+	a: [129]byte;
+	i := len(a);
+	u, neg = is_integer_negative(u, is_signed, bit_size);
+	b := u64(base);
+	for u >= b {
+		i-=1; a[i] = digits[u % b];
+		u /= b;
+	}
+	i-=1; a[i] = digits[u % b];
+
+	if .Prefix in flags {
+		ok := true;
+		switch base {
+		case  2: i-=1; a[i] = 'b';
+		case  8: i-=1; a[i] = 'o';
+		case 10: i-=1; a[i] = 'd';
+		case 12: i-=1; a[i] = 'z';
+		case 16: i-=1; a[i] = 'x';
+		case: ok = false;
+		}
+		if ok {
+			i-=1; a[i] = '0';
+		}
+	}
+
+	switch {
+	case neg:
+		i-=1; a[i] = '-';
+	case .Plus in flags:
+		i-=1; a[i] = '+';
+	case .Space in flags:
+		i-=1; a[i] = ' ';
+	}
+
+	out := a[i:];
+	copy(buf, out);
+	return string(buf[0:len(out)]);
+}
+
+is_integer_negative_128 :: proc(u: u128, is_signed: bool, bit_size: int) -> (unsigned: u128, neg: bool) {
+	if is_signed {
+		switch bit_size {
+		case 8:
+			i := i8(u);
+			neg = i < 0;
+			u = u128(abs(i128(i)));
+		case 16:
+			i := i16(u);
+			neg = i < 0;
+			u = u128(abs(i128(i)));
+		case 32:
+			i := i32(u);
+			neg = i < 0;
+			u = u128(abs(i128(i)));
+		case 64:
+			i := i64(u);
+			neg = i < 0;
+			u = u128(abs(i128(i)));
+		case 128:
+			i := i128(u);
+			neg = i < 0;
+			u = u128(abs(i128(i)));
+		case:
+			panic("is_integer_negative: Unknown integer size");
+		}
+	}
+	return u, neg;
+}
+
+
+append_bits_128 :: proc(buf: []byte, u: u128, base: int, is_signed: bool, bit_size: int, digits: string, flags: Int_Flags) -> string {
+	if base < 2 || base > MAX_BASE {
+		panic("strconv: illegal base passed to append_bits");
+	}
+
+	neg: bool;
+	a: [140]byte;
+	i := len(a);
+	u, neg = is_integer_negative_128(u, is_signed, bit_size);
+	b := u128(base);
+	for u >= b {
+		i-=1; a[i] = digits[u % b];
+		u /= b;
+	}
+	i-=1; a[i] = digits[u % b];
+
+	if .Prefix in flags {
+		ok := true;
+		switch base {
+		case  2: i-=1; a[i] = 'b';
+		case  8: i-=1; a[i] = 'o';
+		case 10: i-=1; a[i] = 'd';
+		case 12: i-=1; a[i] = 'z';
+		case 16: i-=1; a[i] = 'x';
+		case: ok = false;
+		}
+		if ok {
+			i-=1; a[i] = '0';
+		}
+	}
+
+	switch {
+	case neg:
+		i-=1; a[i] = '-';
+	case .Plus in flags:
+		i-=1; a[i] = '+';
+	case .Space in flags:
+		i-=1; a[i] = ' ';
+	}
+
+	out := a[i:];
+	copy(buf, out);
+	return string(buf[0:len(out)]);
+}

core/strconv/strconv.odin

@@ -0,0 +1,301 @@
+package strconv
+
+import "core:unicode/utf8"
+
+parse_bool :: proc(s: string) -> (result: bool = false, ok: bool) {
+	switch s {
+	case "1", "t", "T", "true", "TRUE", "True":
+		return true, true;
+	case "0", "f", "F", "false", "FALSE", "False":
+		return false, true;
+	}
+	return;
+}
+
+_digit_value :: proc(r: rune) -> int {
+	ri := int(r);
+	v: int = 16;
+	switch r {
+	case '0'..'9': v = ri-'0';
+	case 'a'..'z': v = ri-'a'+10;
+	case 'A'..'Z': v = ri-'A'+10;
+	}
+	return v;
+}
+
+parse_i64 :: proc(s: string) -> i64 {
+	neg := false;
+	if len(s) > 1 {
+		switch s[0] {
+		case '-':
+			neg = true;
+			s = s[1:];
+		case '+':
+			s = s[1:];
+		}
+	}
+
+
+	base: i64 = 10;
+	if len(s) > 2 && s[0] == '0' {
+		switch s[1] {
+		case 'b': base =  2;  s = s[2:];
+		case 'o': base =  8;  s = s[2:];
+		case 'd': base = 10;  s = s[2:];
+		case 'z': base = 12;  s = s[2:];
+		case 'x': base = 16;  s = s[2:];
+		}
+	}
+
+
+	value: i64;
+	for r in s {
+		if r == '_' {
+			continue;
+		}
+
+		v := i64(_digit_value(r));
+		if v >= base {
+			break;
+		}
+		value *= base;
+		value += v;
+	}
+
+	if neg do return -value;
+	return value;
+}
+
+parse_u64 :: proc(s: string) -> u64 {
+	neg := false;
+	if len(s) > 1 && s[0] == '+' {
+		s = s[1:];
+	}
+
+
+	base := u64(10);
+	if len(s) > 2 && s[0] == '0' {
+		switch s[1] {
+		case 'b': base =  2;  s = s[2:];
+		case 'o': base =  8;  s = s[2:];
+		case 'd': base = 10;  s = s[2:];
+		case 'z': base = 12;  s = s[2:];
+		case 'x': base = 16;  s = s[2:];
+		}
+	}
+
+
+	value: u64;
+	for r in s {
+		if r == '_' do continue;
+		v := u64(_digit_value(r));
+		if v >= base do break;
+		value *= base;
+		value += u64(v);
+	}
+
+	if neg do return -value;
+	return value;
+}
+
+
+parse_int :: proc(s: string) -> int {
+	return int(parse_i64(s));
+}
+parse_uint :: proc(s: string, base: int) -> uint {
+	return uint(parse_u64(s));
+}
+
+parse_f32 :: proc(s: string) -> f32 {
+	return f32(parse_f64(s));
+}
+
+
+parse_f64 :: proc(s: string) -> f64 {
+	if s == "" {
+		return 0;
+	}
+	i := 0;
+
+	sign: f64 = 1;
+	switch s[i] {
+	case '-': i += 1; sign = -1;
+	case '+': i += 1;
+	}
+
+	value: f64 = 0;
+	for ; i < len(s); i += 1 {
+		r := rune(s[i]);
+		if r == '_' do continue;
+
+		v := _digit_value(r);
+		if v >= 10 do break;
+		value *= 10;
+		value += f64(v);
+	}
+
+	if i < len(s) && s[i] == '.' {
+		pow10: f64 = 10;
+		i += 1;
+
+		for ; i < len(s); i += 1 {
+			r := rune(s[i]);
+			if r == '_' do continue;
+
+			v := _digit_value(r);
+			if v >= 10 do break;
+			value += f64(v)/pow10;
+			pow10 *= 10;
+		}
+	}
+
+	frac := false;
+	scale: f64 = 1;
+
+	if i < len(s) && (s[i] == 'e' || s[i] == 'E') {
+		i += 1;
+
+		if i < len(s) {
+			switch s[i] {
+			case '-': i += 1; frac = true;
+			case '+': i += 1;
+			}
+
+			exp: u32 = 0;
+			for ; i < len(s); i += 1 {
+				r := rune(s[i]);
+				if r == '_' do continue;
+
+				d := u32(_digit_value(r));
+				if d >= 10 do break;
+				exp = exp * 10 + d;
+			}
+			if exp > 308 { exp = 308; }
+
+			for exp >= 50 { scale *= 1e50; exp -= 50; }
+			for exp >=  8 { scale *=  1e8; exp -=  8; }
+			for exp >   0 { scale *=   10; exp -=  1; }
+		}
+	}
+
+	if frac do return sign * (value/scale);
+	return sign * (value*scale);
+}
+
+
+append_bool :: proc(buf: []byte, b: bool) -> string {
+	n := 0;
+	if b do n = copy(buf, cast([]byte)"true");
+	else do n = copy(buf, cast([]byte)"false");
+	return string(buf[:n]);
+}
+
+append_uint :: proc(buf: []byte, u: u64, base: int) -> string {
+	return append_bits(buf, u64(u), base, false, 8*size_of(uint), digits, nil);
+}
+append_int :: proc(buf: []byte, i: i64, base: int) -> string {
+	return append_bits(buf, u64(i), base, true, 8*size_of(int), digits, nil);
+}
+
+itoa :: proc(buf: []byte, i: int) -> string {
+	return append_int(buf, i64(i), 10);
+}
+atoi :: proc(s: string) -> int {
+	return parse_int(s);
+}
+
+append_float :: proc(buf: []byte, f: f64, fmt: byte, prec, bit_size: int) -> string {
+	return string(generic_ftoa(buf, f, fmt, prec, bit_size));
+}
+
+
+quote :: proc(buf: []byte, s: string) -> string {
+	write_byte :: inline proc(buf: []byte, i: ^int, bytes: ..byte) {
+		if i^ >= len(buf) do return;
+		n := copy(buf[i^:], bytes[:]);
+		i^ += n;
+	}
+
+	if buf == nil {
+		return "";
+	}
+
+	c :: '"';
+	i := 0;
+
+	write_byte(buf, &i, c);
+	for width := 0; len(s) > 0; s = s[width:] {
+		r := rune(s[0]);
+		width = 1;
+		if r >= utf8.RUNE_SELF {
+			r, width = utf8.decode_rune_in_string(s);
+		}
+		if width == 1 && r == utf8.RUNE_ERROR {
+			write_byte(buf, &i, '\\', 'x');
+			write_byte(buf, &i, digits[s[0]>>4]);
+			write_byte(buf, &i, digits[s[0]&0xf]);
+		}
+		if i < len(buf) {
+			x := quote_rune(buf[i:], r);
+			i += len(x);
+		}
+	}
+	write_byte(buf, &i, c);
+	return string(buf[:i]);
+}
+
+quote_rune :: proc(buf: []byte, r: rune) -> string {
+	write_byte :: inline proc(buf: []byte, i: ^int, bytes: ..byte) {
+		if i^ < len(buf) {
+			n := copy(buf[i^:], bytes[:]);
+			i^ += n;
+		}
+	}
+	write_string :: inline proc(buf: []byte, i: ^int, s: string) {
+		if i^ < len(buf) {
+			n := copy(buf[i^:], cast([]byte)s);
+			i^ += n;
+		}
+	}
+	write_rune :: inline proc(buf: []byte, i: ^int, r: rune) {
+		if i^ < len(buf) {
+			b, w := utf8.encode_rune(r);
+			n := copy(buf[i^:], b[:w]);
+			i^ += n;
+		}
+	}
+
+	if buf == nil {
+		return "";
+	}
+
+	i := 0;
+	write_byte(buf, &i, '\'');
+
+	switch r {
+	case '\a': write_string(buf, &i, "\\a");
+	case '\b': write_string(buf, &i, "\\b");
+	case '\e': write_string(buf, &i, "\\e");
+	case '\f': write_string(buf, &i, "\\f");
+	case '\n': write_string(buf, &i, "\\n");
+	case '\r': write_string(buf, &i, "\\r");
+	case '\t': write_string(buf, &i, "\\t");
+	case '\v': write_string(buf, &i, "\\v");
+	case:
+		if r < 32 {
+			write_string(buf, &i, "\\x");
+			b: [2]byte;
+			s := append_bits(b[:], u64(r), 16, true, 64, digits, nil);
+			switch len(s) {
+			case 0: write_string(buf, &i, "00");
+			case 1: write_rune(buf, &i, '0');
+			case 2: write_string(buf, &i, s);
+			}
+		} else {
+			write_rune(buf, &i, r);
+		}
+	}
+	write_byte(buf, &i, '\'');
+
+	return string(buf[:i]);
+}

core/strings/builder.odin

@@ -0,0 +1,213 @@
+package strings
+
+import "core:mem"
+import "core:unicode/utf8"
+import "core:strconv"
+
+Builder :: struct {
+	buf: [dynamic]byte,
+}
+
+make_builder :: proc(allocator := context.allocator) -> Builder {
+	return Builder{make([dynamic]byte, allocator)};
+}
+
+destroy_builder :: proc(b: ^Builder) {
+	delete(b.buf);
+	clear(&b.buf);
+}
+
+grow_builder :: proc(b: ^Builder, cap: int) {
+	reserve(&b.buf, cap);
+}
+
+builder_from_slice :: proc(backing: []byte) -> Builder {
+	s := transmute(mem.Raw_Slice)backing;
+	d := mem.Raw_Dynamic_Array{
+		data = s.data,
+		len  = 0,
+		cap  = s.len,
+		allocator = mem.nil_allocator(),
+	};
+	return transmute(Builder)d;
+}
+to_string :: proc(b: Builder) -> string {
+	return string(b.buf[:]);
+}
+
+builder_len :: proc(b: Builder) -> int {
+	return len(b.buf);
+}
+builder_cap :: proc(b: Builder) -> int {
+	return cap(b.buf);
+}
+
+write_byte :: proc(b: ^Builder, x: byte) {
+	append(&b.buf, x);
+}
+
+write_rune :: proc(b: ^Builder, r: rune) -> int {
+	if r < utf8.RUNE_SELF {
+		write_byte(b, byte(r));
+		return 1;
+	}
+
+	s, n := utf8.encode_rune(r);
+	write_bytes(b, s[:n]);
+	return n;
+}
+
+write_string :: proc(b: ^Builder, s: string) {
+	write_bytes(b, cast([]byte)s);
+}
+
+write_bytes :: proc(b: ^Builder, x: []byte) {
+	append(&b.buf, ..x);
+}
+
+@(private, static)
+DIGITS_LOWER := "0123456789abcdefx";
+
+write_quoted_string :: proc(b: ^Builder, s: string, quote: byte = '"') {
+	write_byte(b, quote);
+	for width := 0; len(s) > 0; s = s[width:] {
+		r := rune(s[0]);
+		width = 1;
+		if r >= utf8.RUNE_SELF {
+			r, width = utf8.decode_rune_in_string(s);
+		}
+		if width == 1 && r == utf8.RUNE_ERROR {
+			write_byte(b, '\\');
+			write_byte(b, 'x');
+			write_byte(b, DIGITS_LOWER[s[0]>>4]);
+			write_byte(b, DIGITS_LOWER[s[0]&0xf]);
+			continue;
+		}
+
+		write_escaped_rune(b, r, quote);
+
+	}
+	write_byte(b, quote);
+}
+
+
+write_encoded_rune :: proc(b: ^Builder, r: rune, write_quote := true) {
+	if write_quote do write_byte(b, '\'');
+	switch r {
+	case '\a': write_string(b, `\a"`);
+	case '\b': write_string(b, `\b"`);
+	case '\e': write_string(b, `\e"`);
+	case '\f': write_string(b, `\f"`);
+	case '\n': write_string(b, `\n"`);
+	case '\r': write_string(b, `\r"`);
+	case '\t': write_string(b, `\t"`);
+	case '\v': write_string(b, `\v"`);
+	case:
+		if r < 32 {
+			write_string(b, `\x`);
+			buf: [2]byte;
+			s := strconv.append_bits(buf[:], u64(r), 16, true, 64, strconv.digits, nil);
+			switch len(s) {
+			case 0: write_string(b, "00");
+			case 1: write_byte(b, '0');
+			case 2: write_string(b, s);
+			}
+		} else {
+			write_rune(b, r);
+		}
+
+	}
+	if write_quote do write_byte(b, '\'');
+}
+
+
+write_escaped_rune :: proc(b: ^Builder, r: rune, quote: byte, html_safe := false) {
+	is_printable :: proc(r: rune) -> bool {
+		if r <= 0xff {
+			switch r {
+			case 0x20..0x7e:
+				return true;
+			case 0xa1..0xff: // ¡ through ÿ except for the soft hyphen
+				return r != 0xad; //
+			}
+		}
+
+		// TODO(bill): A proper unicode library will be needed!
+		return false;
+	}
+
+	if html_safe {
+		switch r {
+		case '<', '>', '&':
+			write_byte(b, '\\');
+			write_byte(b, 'u');
+			for s := 12; s >= 0; s -= 4 {
+				write_byte(b, DIGITS_LOWER[r>>uint(s) & 0xf]);
+			}
+			return;
+		}
+	}
+
+	if r == rune(quote) || r == '\\' {
+		write_byte(b, '\\');
+		write_byte(b, byte(r));
+		return;
+	} else if is_printable(r) {
+		write_encoded_rune(b, r, false);
+		return;
+	}
+	switch r {
+	case '\a': write_string(b, `\a`);
+	case '\b': write_string(b, `\b`);
+	case '\e': write_string(b, `\e`);
+	case '\f': write_string(b, `\f`);
+	case '\n': write_string(b, `\n`);
+	case '\r': write_string(b, `\r`);
+	case '\t': write_string(b, `\t`);
+	case '\v': write_string(b, `\v`);
+	case:
+		switch {
+		case r < ' ':
+			write_byte(b, '\\');
+			write_byte(b, 'x');
+			write_byte(b, DIGITS_LOWER[byte(r)>>4]);
+			write_byte(b, DIGITS_LOWER[byte(r)&0xf]);
+
+		case r > utf8.MAX_RUNE:
+			r = 0xfffd;
+			fallthrough;
+		case r < 0x10000:
+			write_byte(b, '\\');
+			write_byte(b, 'u');
+			for s := 12; s >= 0; s -= 4 {
+				write_byte(b, DIGITS_LOWER[r>>uint(s) & 0xf]);
+			}
+		case:
+			write_byte(b, '\\');
+			write_byte(b, 'U');
+			for s := 28; s >= 0; s -= 4 {
+				write_byte(b, DIGITS_LOWER[r>>uint(s) & 0xf]);
+			}
+		}
+	}
+}
+
+
+write_u64 :: proc(b: ^Builder, i: u64, base: int = 10) {
+	buf: [32]byte;
+	s := strconv.append_bits(buf[:], u64(i), base, false, 64, strconv.digits, nil);
+	write_string(b, s);
+}
+write_i64 :: proc(b: ^Builder, i: i64, base: int = 10) {
+	buf: [32]byte;
+	s := strconv.append_bits(buf[:], u64(i), base, true, 64, strconv.digits, nil);
+	write_string(b, s);
+}
+
+write_uint :: proc(b: ^Builder, i: uint, base: int = 10) {
+	write_u64(b, u64(i), base);
+}
+write_int :: proc(b: ^Builder, i: int, base: int = 10) {
+	write_i64(b, i64(i), base);
+}
+

core/strings/strings.odin

@@ -0,0 +1,693 @@
+package strings
+
+import "core:mem"
+import "core:unicode/utf8"
+
+clone :: proc(s: string, allocator := context.allocator) -> string {
+	c := make([]byte, len(s)+1, allocator);
+	copy(c, cast([]byte)s);
+	c[len(s)] = 0;
+	return string(c[:len(s)]);
+}
+
+clone_to_cstring :: proc(s: string, allocator := context.allocator) -> cstring {
+	c := make([]byte, len(s)+1, allocator);
+	copy(c, cast([]byte)s);
+	c[len(s)] = 0;
+	return cstring(&c[0]);
+}
+
+@(deprecated="Please use 'strings.clone'")
+new_string :: proc(s: string, allocator := context.allocator) -> string {
+	c := make([]byte, len(s)+1, allocator);
+	copy(c, cast([]byte)s);
+	c[len(s)] = 0;
+	return string(c[:len(s)]);
+}
+
+@(deprecated="Please use 'strings.clone_to_cstring'")
+new_cstring :: proc(s: string, allocator := context.allocator) -> cstring {
+	c := make([]byte, len(s)+1, allocator);
+	copy(c, cast([]byte)s);
+	c[len(s)] = 0;
+	return cstring(&c[0]);
+}
+
+@(deprecated="Please use a standard cast for cstring to string")
+to_odin_string :: proc(str: cstring) -> string {
+	return string(str);
+}
+
+string_from_ptr :: proc(ptr: ^byte, len: int) -> string {
+	return transmute(string)mem.Raw_String{ptr, len};
+}
+
+compare :: proc(lhs, rhs: string) -> int {
+	return mem.compare(cast([]byte)lhs, cast([]byte)rhs);
+}
+
+contains_rune :: proc(s: string, r: rune) -> int {
+	for c, offset in s {
+		if c == r do return offset;
+	}
+	return -1;
+}
+
+contains :: proc(s, substr: string) -> bool {
+	return index(s, substr) >= 0;
+}
+
+contains_any :: proc(s, chars: string) -> bool {
+	return index_any(s, chars) >= 0;
+}
+
+
+rune_count :: proc(s: string) -> int {
+	return utf8.rune_count_in_string(s);
+}
+
+
+equal_fold :: proc(s, t: string) -> bool {
+	loop: for s != "" && t != "" {
+		sr, tr: rune;
+		if s[0] < utf8.RUNE_SELF {
+			sr, s = rune(s[0]), s[1:];
+		} else {
+			r, size := utf8.decode_rune_in_string(s);
+			sr, s = r, s[size:];
+		}
+		if t[0] < utf8.RUNE_SELF {
+			tr, t = rune(t[0]), t[1:];
+		} else {
+			r, size := utf8.decode_rune_in_string(t);
+			tr, t = r, t[size:];
+		}
+
+		if tr == sr { // easy case
+			continue loop;
+		}
+
+		if tr < sr {
+			tr, sr = sr, tr;
+		}
+
+		if tr < utf8.RUNE_SELF {
+			switch sr {
+			case 'A'..'Z':
+				if tr == (sr+'a')-'A' {
+					continue loop;
+				}
+			}
+			return false;
+		}
+
+		// TODO(bill): Unicode folding
+
+		return false;
+	}
+
+	return s == t;
+}
+
+has_prefix :: proc(s, prefix: string) -> bool {
+	return len(s) >= len(prefix) && s[0:len(prefix)] == prefix;
+}
+
+has_suffix :: proc(s, suffix: string) -> bool {
+	return len(s) >= len(suffix) && s[len(s)-len(suffix):] == suffix;
+}
+
+
+join :: proc(a: []string, sep: string, allocator := context.allocator) -> string {
+	if len(a) == 0 {
+		return "";
+	}
+
+	n := len(sep) * (len(a) - 1);
+	for s in a {
+		n += len(s);
+	}
+
+	b := make([]byte, n, allocator);
+	i := copy(b, cast([]byte)a[0]);
+	for s in a[1:] {
+		i += copy(b[i:], cast([]byte)sep);
+		i += copy(b[i:], cast([]byte)s);
+	}
+	return string(b);
+}
+
+concatenate :: proc(a: []string, allocator := context.allocator) -> string {
+	if len(a) == 0 {
+		return "";
+	}
+
+	n := 0;
+	for s in a {
+		n += len(s);
+	}
+	b := make([]byte, n, allocator);
+	i := 0;
+	for s in a {
+		i += copy(b[i:], cast([]byte)s);
+	}
+	return string(b);
+}
+
+index_byte :: proc(s: string, c: byte) -> int {
+	for i := 0; i < len(s); i += 1 {
+		if s[i] == c do return i;
+	}
+	return -1;
+}
+
+// Returns i1 if c is not present
+last_index_byte :: proc(s: string, c: byte) -> int {
+	for i := len(s)-1; i >= 0; i -= 1 {
+		if s[i] == c do return i;
+	}
+	return -1;
+}
+
+index :: proc(s, substr: string) -> int {
+	n := len(substr);
+	switch {
+	case n == 0:
+		return 0;
+	case n == 1:
+		return index_byte(s, substr[0]);
+	case n == len(s):
+		if s == substr {
+			return 0;
+		}
+		return -1;
+	case n > len(s):
+		return -1;
+	}
+
+	for i := 0; i < len(s)-n+1; i += 1 {
+		x := s[i:i+n];
+		if x == substr {
+			return i;
+		}
+	}
+	return -1;
+}
+
+index_any :: proc(s, chars: string) -> int {
+	if chars == "" {
+		return -1;
+	}
+
+	// TODO(bill): Optimize
+	for r, i in s {
+		for c in chars {
+			if r == c {
+				return i;
+			}
+		}
+	}
+	return -1;
+}
+
+last_index_any :: proc(s, chars: string) -> int {
+	if chars == "" {
+		return -1;
+	}
+
+	for i := len(s); i > 0;  {
+		r, w := utf8.decode_last_rune_in_string(s[:i]);
+		i -= w;
+		for c in chars {
+			if r == c {
+				return i;
+			}
+		}
+	}
+	return -1;
+}
+
+count :: proc(s, substr: string) -> int {
+	if len(substr) == 0 { // special case
+		return rune_count(s) + 1;
+	}
+	if len(substr) == 1 {
+		c := substr[0];
+		switch len(s) {
+		case 0:
+			return 0;
+		case 1:
+			return int(s[0] == c);
+		}
+		n := 0;
+		for i := 0; i < len(s); i += 1 {
+			if s[i] == c {
+				n += 1;
+			}
+		}
+		return n;
+	}
+
+	// TODO(bill): Use a non-brute for approach
+	n := 0;
+	for {
+		i := index(s, substr);
+		if i == -1 {
+			return n;
+		}
+		n += 1;
+		s = s[i+len(substr):];
+	}
+	return n;
+}
+
+
+repeat :: proc(s: string, count: int, allocator := context.allocator) -> string {
+	if count < 0 {
+		panic("strings: negative repeat count");
+	} else if count > 0 && (len(s)*count)/count != len(s) {
+		panic("strings: repeat count will cause an overflow");
+	}
+
+	b := make([]byte, len(s)*count, allocator);
+	i := copy(b, cast([]byte)s);
+	for i < len(b) { // 2^N trick to reduce the need to copy
+		copy(b[i:], b[:i]);
+		i *= 2;
+	}
+	return string(b);
+}
+
+replace_all :: proc(s, old, new: string, allocator := context.allocator) -> (output: string, was_allocation: bool) {
+	return replace(s, old, new, -1, allocator);
+}
+
+// if n < 0, no limit on the number of replacements
+replace :: proc(s, old, new: string, n: int, allocator := context.allocator) -> (output: string, was_allocation: bool) {
+	if old == new || n == 0 {
+		was_allocation = false;
+		output = s;
+		return;
+	}
+
+	if m := count(s, old); m == 0 {
+		was_allocation = false;
+		output = s;
+		return;
+	} else if n < 0 || m < n {
+		n = m;
+	}
+
+
+	t := make([]byte, len(s) + n*(len(new) - len(old)), allocator);
+	was_allocation = true;
+
+	w := 0;
+	start := 0;
+	for i := 0; i < n; i += 1 {
+		j := start;
+		if len(old) == 0 {
+			if i > 0 {
+				_, width := utf8.decode_rune_in_string(s[start:]);
+				j += width;
+			}
+		} else {
+			j += index(s[start:], old);
+		}
+		w += copy(t[w:], cast([]byte)s[start:j]);
+		w += copy(t[w:], cast([]byte)new);
+		start = j + len(old);
+	}
+	w += copy(t[w:], cast([]byte)s[start:]);
+	output = string(t[0:w]);
+	return;
+}
+
+is_ascii_space :: proc(r: rune) -> bool {
+	switch r {
+	case '\t', '\n', '\v', '\f', '\r', ' ':
+		return true;
+	}
+	return false;
+}
+
+is_space :: proc(r: rune) -> bool {
+	if r < 0x2000 {
+		switch r {
+		case '\t', '\n', '\v', '\f', '\r', ' ', 0x85, 0xa0, 0x1680:
+			return true;
+		}
+	} else {
+		if r <= 0x200a {
+			return true;
+		}
+		switch r {
+		case 0x2028, 0x2029, 0x202f, 0x205f, 0x3000:
+			return true;
+		}
+	}
+	return false;
+}
+
+is_null :: proc(r: rune) -> bool {
+	return r == 0x0000;
+}
+
+index_proc :: proc(s: string, p: proc(rune) -> bool, truth := true) -> int {
+	for r, i in s {
+		if p(r) == truth {
+			return i;
+		}
+	}
+	return -1;
+}
+
+index_proc_with_state :: proc(s: string, p: proc(rawptr, rune) -> bool, state: rawptr, truth := true) -> int {
+	for r, i in s {
+		if p(state, r) == truth {
+			return i;
+		}
+	}
+	return -1;
+}
+
+last_index_proc :: proc(s: string, p: proc(rune) -> bool, truth := true) -> int {
+	// TODO(bill): Probably use Rabin-Karp Search
+	for i := len(s); i > 0; {
+		r, size := utf8.decode_last_rune_in_string(s[:i]);
+		i -= size;
+		if p(r) == truth {
+			return i;
+		}
+	}
+	return -1;
+}
+
+last_index_proc_with_state :: proc(s: string, p: proc(rawptr, rune) -> bool, state: rawptr, truth := true) -> int {
+	// TODO(bill): Probably use Rabin-Karp Search
+	for i := len(s); i > 0; {
+		r, size := utf8.decode_last_rune_in_string(s[:i]);
+		i -= size;
+		if p(state, r) == truth {
+			return i;
+		}
+	}
+	return -1;
+}
+
+trim_left_proc :: proc(s: string, p: proc(rune) -> bool) -> string {
+	i := index_proc(s, p, false);
+	if i == -1 {
+		return "";
+	}
+	return s[i:];
+}
+
+
+index_rune :: proc(s: string, r: rune) -> int {
+	switch {
+	case 0 <= r && r < utf8.RUNE_SELF:
+		return index_byte(s, byte(r));
+
+	case r == utf8.RUNE_ERROR:
+		for c, i in s {
+			if c == utf8.RUNE_ERROR {
+				return i;
+			}
+		}
+		return -1;
+
+	case !utf8.valid_rune(r):
+		return -1;
+	}
+
+	b, w := utf8.encode_rune(r);
+	return index(s, string(b[:w]));
+}
+
+
+trim_left_proc_with_state :: proc(s: string, p: proc(rawptr, rune) -> bool, state: rawptr) -> string {
+	i := index_proc_with_state(s, p, state, false);
+	if i == -1 {
+		return "";
+	}
+	return s[i:];
+}
+
+trim_right_proc :: proc(s: string, p: proc(rune) -> bool) -> string {
+	i := last_index_proc(s, p, false);
+	if i >= 0 && s[i] >= utf8.RUNE_SELF {
+		_, w := utf8.decode_rune_in_string(s[i:]);
+		i += w;
+	} else {
+		i += 1;
+	}
+	return s[0:i];
+}
+
+trim_right_proc_with_state :: proc(s: string, p: proc(rawptr, rune) -> bool, state: rawptr) -> string {
+	i := last_index_proc_with_state(s, p, state, false);
+	if i >= 0 && s[i] >= utf8.RUNE_SELF {
+		_, w := utf8.decode_rune_in_string(s[i:]);
+		i += w;
+	} else {
+		i += 1;
+	}
+	return s[0:i];
+}
+
+
+is_in_cutset :: proc(state: rawptr, r: rune) -> bool {
+	if state == nil {
+		return false;
+	}
+	cutset := (^string)(state)^;
+	for c in cutset {
+		if r == c {
+			return true;
+		}
+	}
+	return false;
+}
+
+
+trim_left :: proc(s: string, cutset: string) -> string {
+	if s == "" || cutset == "" {
+		return s;
+	}
+	return trim_left_proc_with_state(s, is_in_cutset, &cutset);
+}
+
+trim_right :: proc(s: string, cutset: string) -> string {
+	if s == "" || cutset == "" {
+		return s;
+	}
+	return trim_right_proc_with_state(s, is_in_cutset, &cutset);
+}
+
+trim :: proc(s: string, cutset: string) -> string {
+	return trim_right(trim_left(s, cutset), cutset);
+}
+
+trim_left_space :: proc(s: string) -> string {
+	return trim_left_proc(s, is_space);
+}
+
+trim_right_space :: proc(s: string) -> string {
+	return trim_right_proc(s, is_space);
+}
+
+trim_space :: proc(s: string) -> string {
+	return trim_right_space(trim_left_space(s));
+}
+
+trim_left_null :: proc(s: string) -> string {
+	return trim_left_proc(s, is_null);
+}
+
+trim_right_null :: proc(s: string) -> string {
+	return trim_right_proc(s, is_null);
+}
+
+trim_null :: proc(s: string) -> string {
+	return trim_right_null(trim_left_null(s));
+}
+
+// scrub scruvs invalid utf-8 characters and replaces them with the replacement string
+// Adjacent invalid bytes are only replaced once
+scrub :: proc(str: string, replacement: string, allocator := context.allocator) -> string {
+	b := make_builder(allocator);;
+	grow_builder(&b, len(str));
+
+	has_error := false;
+	cursor := 0;
+	origin := str;
+
+	for len(str) > 0 {
+		r, w := utf8.decode_rune_in_string(str);
+
+		if r == utf8.RUNE_ERROR {
+			if !has_error {
+				has_error = true;
+				write_string(&b, origin[:cursor]);
+			}
+		} else if has_error {
+			has_error = false;
+			write_string(&b, replacement);
+
+			origin = origin[cursor:];
+			cursor = 0;
+		}
+
+		cursor += w;
+		str = str[w:];
+	}
+
+	return to_string(b);
+}
+
+
+reverse :: proc(str: string, allocator := context.allocator) -> string {
+	n := len(str);
+	buf := make([]byte, n);
+	i := 0;
+
+	for len(str) > 0 {
+		_, w := utf8.decode_rune_in_string(str);
+		copy(buf[i:], cast([]byte)str[:w]);
+		str = str[w:];
+	}
+	return string(buf);
+}
+
+expand_tabs :: proc(str: string, tab_size: int, allocator := context.allocator) -> string {
+	if tab_size <= 0 {
+		panic("tab size must be positive");
+	}
+
+	if str == "" {
+		return "";
+	}
+
+	b := make_builder(allocator);
+
+	column: int;
+
+	for len(str) > 0 {
+		r, w := utf8.decode_rune_in_string(str);
+
+		if r == '\t' {
+			expand := tab_size - column%tab_size;
+
+			for i := 0; i < expand; i += 1 {
+				write_byte(&b, ' ');
+			}
+
+			column += expand;
+		} else {
+			if r == '\n' {
+				column = 0;
+			} else {
+				column += w;
+			}
+
+			write_rune(&b, r);
+		}
+
+		str = str[w:];
+	}
+
+	return to_string(b);
+}
+
+
+partition :: proc(str, sep: string) -> (head, match, tail: string) {
+	i := index(str, sep);
+	if i == -1 {
+		head = str;
+		return;
+	}
+
+	head = str[:i];
+	match = str[i:i+len(sep)];
+	tail = str[i+len(sep):];
+	return;
+}
+
+center_justify :: centre_justify; // NOTE(bill): Because Americans exist
+
+// centre_justify returns a string with a pad string at boths sides if the str's rune length is smaller than length
+centre_justify :: proc(str: string, length: int, pad: string, allocator := context.allocator) -> string {
+	n := rune_count(str);
+	if n >= length || pad == "" {
+		return clone(str, allocator);
+	}
+
+	remains := length-1;
+	pad_len := rune_count(pad);
+
+	b := make_builder(allocator);
+	grow_builder(&b, len(str) + (remains/pad_len + 1)*len(pad));
+
+	write_pad_string(&b, pad, pad_len, remains/2);
+	write_string(&b, str);
+	write_pad_string(&b, pad, pad_len, (remains+1)/2);
+
+	return to_string(b);
+}
+
+// left_justify returns a string with a pad string at left side if the str's rune length is smaller than length
+left_justify :: proc(str: string, length: int, pad: string, allocator := context.allocator) -> string {
+	n := rune_count(str);
+	if n >= length || pad == "" {
+		return clone(str, allocator);
+	}
+
+	remains := length-1;
+	pad_len := rune_count(pad);
+
+	b := make_builder(allocator);
+	grow_builder(&b, len(str) + (remains/pad_len + 1)*len(pad));
+
+	write_string(&b, str);
+	write_pad_string(&b, pad, pad_len, remains);
+
+	return to_string(b);
+}
+
+// right_justify returns a string with a pad string at right side if the str's rune length is smaller than length
+right_justify :: proc(str: string, length: int, pad: string, allocator := context.allocator) -> string {
+	n := rune_count(str);
+	if n >= length || pad == "" {
+		return clone(str, allocator);
+	}
+
+	remains := length-1;
+	pad_len := rune_count(pad);
+
+	b := make_builder(allocator);
+	grow_builder(&b, len(str) + (remains/pad_len + 1)*len(pad));
+
+	write_pad_string(&b, pad, pad_len, remains);
+	write_string(&b, str);
+
+	return to_string(b);
+}
+
+
+@private
+write_pad_string :: proc(b: ^Builder, pad: string, pad_len, remains: int) {
+	repeats := remains / pad_len;
+
+	for i := 0; i < repeats; i += 1 {
+		write_string(b, pad);
+	}
+
+	remains = remains % pad_len;
+
+	if remains != 0 do for i := 0; i < remains; i += 1 {
+		r, w := utf8.decode_rune_in_string(pad);
+		write_rune(b, r);
+		pad = pad[w:];
+	}
+}

core/sync.odin

@@ -1,91 +0,0 @@
-#import win32 "sys/windows.odin" when ODIN_OS == "windows";
-#import "atomic.odin";
-
-Semaphore :: struct {
-	handle: win32.HANDLE,
-}
-
-Mutex :: struct {
-	semaphore: Semaphore,
-	counter:   i32,
-	owner:     i32,
-	recursion: i32,
-}
-
-current_thread_id :: proc() -> i32 {
-	return cast(i32)win32.GetCurrentThreadId();
-}
-
-semaphore_init :: proc(s: ^Semaphore) {
-	s.handle = win32.CreateSemaphoreA(nil, 0, 1<<31-1, nil);
-}
-
-semaphore_destroy :: proc(s: ^Semaphore) {
-	win32.CloseHandle(s.handle);
-}
-
-semaphore_post :: proc(s: ^Semaphore, count: int) {
-	win32.ReleaseSemaphore(s.handle, cast(i32)count, nil);
-}
-
-semaphore_release :: proc(s: ^Semaphore) #inline { semaphore_post(s, 1); }
-
-semaphore_wait :: proc(s: ^Semaphore) {
-	win32.WaitForSingleObject(s.handle, win32.INFINITE);
-}
-
-
-mutex_init :: proc(m: ^Mutex) {
-	atomic.store(^m.counter, 0);
-	atomic.store(^m.owner, current_thread_id());
-	semaphore_init(^m.semaphore);
-	m.recursion = 0;
-}
-mutex_destroy :: proc(m: ^Mutex) {
-	semaphore_destroy(^m.semaphore);
-}
-mutex_lock :: proc(m: ^Mutex) {
-	thread_id := current_thread_id();
-	if atomic.fetch_add(^m.counter, 1) > 0 {
-		if thread_id != atomic.load(^m.owner) {
-			semaphore_wait(^m.semaphore);
-		}
-	}
-	atomic.store(^m.owner, thread_id);
-	m.recursion += 1;
-}
-mutex_try_lock :: proc(m: ^Mutex) -> bool {
-	thread_id := current_thread_id();
-	if atomic.load(^m.owner) == thread_id {
-		atomic.fetch_add(^m.counter, 1);
-	} else {
-		expected: i32 = 0;
-		if atomic.load(^m.counter) != 0 {
-			return false;
-		}
-		if atomic.compare_exchange(^m.counter, expected, 1) == 0 {
-			return false;
-		}
-		atomic.store(^m.owner, thread_id);
-	}
-	m.recursion += 1;
-	return true;
-}
-mutex_unlock :: proc(m: ^Mutex) {
-	recursion: i32;
-	thread_id := current_thread_id();
-	assert(thread_id == atomic.load(^m.owner));
-
-	m.recursion -= 1;
-	recursion = m.recursion;
-	if recursion == 0 {
-		atomic.store(^m.owner, thread_id);
-	}
-
-	if atomic.fetch_add(^m.counter, -1) > 1 {
-		if recursion == 0 {
-			semaphore_release(^m.semaphore);
-		}
-	}
-}
-

core/sync/atomic.odin

@@ -0,0 +1,211 @@
+package sync
+
+import "intrinsics"
+
+Ordering :: enum {
+	Relaxed, // Monotonic
+	Release,
+	Acquire,
+	Acquire_Release,
+	Sequentially_Consistent,
+}
+
+strongest_failure_ordering :: inline proc "contextless" (order: Ordering) -> Ordering {
+	#complete switch order {
+	case .Relaxed:                 return .Relaxed;
+	case .Release:                 return .Relaxed;
+	case .Acquire:                 return .Acquire;
+	case .Acquire_Release:         return .Acquire;
+	case .Sequentially_Consistent: return .Sequentially_Consistent;
+	}
+	return .Relaxed;
+}
+
+fence :: inline proc "contextless" ($order: Ordering) {
+	#complete switch order {
+	case .Relaxed:                 panic("there is no such thing as a relaxed fence");
+	case .Release:                 intrinsics.atomic_fence_rel();
+	case .Acquire:                 intrinsics.atomic_fence_acq();
+	case .Acquire_Release:         intrinsics.atomic_fence_acqrel();
+	case .Sequentially_Consistent: intrinsics.atomic_fence();
+	case: panic("unknown order");
+	}
+}
+
+
+atomic_store :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) {
+	#complete switch order {
+	case .Relaxed:                 intrinsics.atomic_store_relaxed(dst, val);
+	case .Release:                 intrinsics.atomic_store_rel(dst, val);
+	case .Sequentially_Consistent: intrinsics.atomic_store(dst, val);
+	case .Acquire:         panic("there is not such thing as an acquire store");
+	case .Acquire_Release: panic("there is not such thing as an acquire/release store");
+	case: panic("unknown order");
+	}
+}
+
+atomic_load :: inline proc "contextless" (dst: ^$T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_load_relaxed(dst);
+	case .Acquire:                 return intrinsics.atomic_load_acq(dst);
+	case .Sequentially_Consistent: return intrinsics.atomic_load(dst);
+	case .Release:         panic("there is no such thing as a release load");
+	case .Acquire_Release: panic("there is no such thing as an acquire/release load");
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_swap :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_xchg_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_xchg_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_xchg_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_xchg_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_xchg(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_compare_exchange :: inline proc "contextless" (dst: ^$T, old, new: T, $success, $failure: Ordering) -> (val: T, ok: bool) {
+	switch failure {
+	case .Relaxed:
+		switch success {
+		case .Relaxed:                 return intrinsics.atomic_cxchg_relaxed(dst, old, new);
+		case .Acquire:                 return intrinsics.atomic_cxchg_acq_failrelaxed(dst, old, new);
+		case .Acquire_Release:         return intrinsics.atomic_cxchg_acqrel_failrelaxed(dst, old, new);
+		case .Sequentially_Consistent: return intrinsics.atomic_cxchg_failrelaxed(dst, old, new);
+		case .Release:                 return intrinsics.atomic_cxchg_rel(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Acquire:
+		switch success {
+		case .Release:                 return intrinsics.atomic_cxchg_acqrel(dst, old, new);
+		case .Acquire:                 return intrinsics.atomic_cxchg_acq(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Sequentially_Consistent:
+		switch success {
+		case .Sequentially_Consistent: return intrinsics.atomic_cxchg(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Acquire_Release:
+		panic("there is not such thing as an acquire/release failure ordering");
+	case .Release:
+		switch success {
+		case .Acquire:                 return instrinsics.atomic_cxchg_failacq(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	}
+	return T{}, false;
+
+}
+
+atomic_compare_exchange_weak :: inline proc "contextless" (dst: ^$T, old, new: T, $success, $failure: Ordering) -> (val: T, ok: bool) {
+	switch failure {
+	case .Relaxed:
+		switch success {
+		case .Relaxed:                 return intrinsics.atomic_cxchgweak_relaxed(dst, old, new);
+		case .Acquire:                 return intrinsics.atomic_cxchgweak_acq_failrelaxed(dst, old, new);
+		case .Acquire_Release:         return intrinsics.atomic_cxchgweak_acqrel_failrelaxed(dst, old, new);
+		case .Sequentially_Consistent: return intrinsics.atomic_cxchgweak_failrelaxed(dst, old, new);
+		case .Release:                 return intrinsics.atomic_cxchgweak_rel(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Acquire:
+		switch success {
+		case .Release:                 return intrinsics.atomic_cxchgweak_acqrel(dst, old, new);
+		case .Acquire:                 return intrinsics.atomic_cxchgweak_acq(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Sequentially_Consistent:
+		switch success {
+		case .Sequentially_Consistent: return intrinsics.atomic_cxchgweak(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	case .Acquire_Release:
+		panic("there is not such thing as an acquire/release failure ordering");
+	case .Release:
+		switch success {
+		case .Acquire:                 return intrinsics.atomic_cxchgweak_failacq(dst, old, new);
+		case: panic("an unknown ordering combination");
+		}
+	}
+	return T{}, false;
+
+}
+
+
+atomic_add :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_add_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_add_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_add_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_add_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_add(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_sub :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_sub_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_sub_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_sub_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_sub_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_sub(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_and :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_and_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_and_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_and_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_and_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_and(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_nand :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_nand_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_nand_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_nand_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_nand_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_nand(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_or :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_or_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_or_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_or_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_or_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_or(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+
+atomic_xor :: inline proc "contextless" (dst: ^$T, val: T, $order: Ordering) -> T {
+	#complete switch order {
+	case .Relaxed:                 return intrinsics.atomic_xor_relaxed(dst, val);
+	case .Release:                 return intrinsics.atomic_xor_rel(dst, val);
+	case .Acquire:                 return intrinsics.atomic_xor_acq(dst, val);
+	case .Acquire_Release:         return intrinsics.atomic_xor_acqrel(dst, val);
+	case .Sequentially_Consistent: return intrinsics.atomic_xor(dst, val);
+	}
+	panic("unknown order");
+	return T{};
+}
+

core/sync/sync_linux.odin

@@ -0,0 +1,98 @@
+package sync
+
+/*
+
+import "core:atomics"
+import "core:os"
+
+Semaphore :: struct {
+	// _handle: win32.Handle,
+}
+
+Mutex :: struct {
+	_semaphore: Semaphore,
+	_counter:   i32,
+	_owner:     i32,
+	_recursion: i32,
+}
+
+current_thread_id :: proc() -> i32 {
+	return i32(os.current_thread_id());
+}
+
+semaphore_init :: proc(s: ^Semaphore) {
+	// s._handle = win32.CreateSemaphoreA(nil, 0, 1<<31-1, nil);
+}
+
+semaphore_destroy :: proc(s: ^Semaphore) {
+	// win32.CloseHandle(s._handle);
+}
+
+semaphore_post :: proc(s: ^Semaphore, count: int) {
+	// win32.ReleaseSemaphore(s._handle, cast(i32)count, nil);
+}
+
+semaphore_release :: inline proc(s: ^Semaphore) {
+	semaphore_post(s, 1);
+}
+
+semaphore_wait :: proc(s: ^Semaphore) {
+	// win32.WaitForSingleObject(s._handle, win32.INFINITE);
+}
+
+
+mutex_init :: proc(m: ^Mutex) {
+	atomics.store(&m._counter, 0);
+	atomics.store(&m._owner, current_thread_id());
+	semaphore_init(&m._semaphore);
+	m._recursion = 0;
+}
+mutex_destroy :: proc(m: ^Mutex) {
+	semaphore_destroy(&m._semaphore);
+}
+mutex_lock :: proc(m: ^Mutex) {
+	thread_id := current_thread_id();
+	if atomics.fetch_add(&m._counter, 1) > 0 {
+		if thread_id != atomics.load(&m._owner) {
+			semaphore_wait(&m._semaphore);
+		}
+	}
+	atomics.store(&m._owner, thread_id);
+	m._recursion += 1;
+}
+mutex_try_lock :: proc(m: ^Mutex) -> bool {
+	thread_id := current_thread_id();
+	if atomics.load(&m._owner) == thread_id {
+		atomics.fetch_add(&m._counter, 1);
+	} else {
+		expected: i32 = 0;
+		if atomics.load(&m._counter) != 0 {
+			return false;
+		}
+		if atomics.compare_exchange(&m._counter, expected, 1) == 0 {
+			return false;
+		}
+		atomics.store(&m._owner, thread_id);
+	}
+	m._recursion += 1;
+	return true;
+}
+mutex_unlock :: proc(m: ^Mutex) {
+	recursion: i32;
+	thread_id := current_thread_id();
+	assert(thread_id == atomics.load(&m._owner));
+
+	m._recursion -= 1;
+	recursion = m._recursion;
+	if recursion == 0 {
+		atomics.store(&m._owner, thread_id);
+	}
+
+	if atomics.fetch_add(&m._counter, -1) > 1 {
+		if recursion == 0 {
+			semaphore_release(&m._semaphore);
+		}
+	}
+}
+
+*/

core/sync/sync_windows.odin

@@ -0,0 +1,111 @@
+package sync
+
+import "core:sys/win32"
+
+foreign {
+	@(link_name="llvm.x86.sse2.pause")
+	yield_processor :: proc() ---
+}
+
+Semaphore :: struct {
+	_handle: win32.Handle,
+}
+
+Mutex :: struct {
+	_critical_section: win32.Critical_Section,
+}
+
+Condition :: struct {
+	event: win32.Handle,
+}
+
+Ticket_Mutex :: struct {
+	ticket:  u64,
+	serving: u64,
+}
+
+
+current_thread_id :: proc() -> i32 {
+	return i32(win32.get_current_thread_id());
+}
+
+semaphore_init :: proc(s: ^Semaphore) {
+	s._handle = win32.create_semaphore_w(nil, 0, 1<<31-1, nil);
+}
+
+semaphore_destroy :: proc(s: ^Semaphore) {
+	win32.close_handle(s._handle);
+}
+
+semaphore_post :: proc(s: ^Semaphore, count: int) {
+	win32.release_semaphore(s._handle, i32(count), nil);
+}
+
+semaphore_release :: inline proc(s: ^Semaphore) {
+	semaphore_post(s, 1);
+}
+
+semaphore_wait :: proc(s: ^Semaphore) {
+	result := win32.wait_for_single_object(s._handle, win32.INFINITE);
+	assert(result != win32.WAIT_FAILED);
+}
+
+
+mutex_init :: proc(m: ^Mutex, spin_count := 0) {
+	win32.initialize_critical_section_and_spin_count(&m._critical_section, u32(spin_count));
+}
+
+mutex_destroy :: proc(m: ^Mutex) {
+	win32.delete_critical_section(&m._critical_section);
+}
+
+mutex_lock :: proc(m: ^Mutex) {
+	win32.enter_critical_section(&m._critical_section);
+}
+
+mutex_try_lock :: proc(m: ^Mutex) -> bool {
+	return bool(win32.try_enter_critical_section(&m._critical_section));
+}
+
+mutex_unlock :: proc(m: ^Mutex) {
+	win32.leave_critical_section(&m._critical_section);
+}
+
+
+condition_init :: proc(using c: ^Condition) {
+	event = win32.create_event_w(nil, false, false, nil);
+	assert(event != nil);
+}
+
+condition_signal :: proc(using c: ^Condition) {
+	ok := win32.set_event(event);
+	assert(bool(ok));
+}
+
+condition_wait_for :: proc(using c: ^Condition) {
+	result := win32.wait_for_single_object(event, win32.INFINITE);
+	assert(result != win32.WAIT_FAILED);
+}
+
+condition_destroy :: proc(using c: ^Condition) {
+	if event != nil {
+		win32.close_handle(event);
+	}
+}
+
+
+ticket_mutex_init :: proc(m: ^Ticket_Mutex) {
+	atomic_store(&m.ticket,  0, Ordering.Relaxed);
+	atomic_store(&m.serving, 0, Ordering.Relaxed);
+}
+
+ticket_mutex_lock :: inline proc(m: ^Ticket_Mutex) {
+	ticket := atomic_add(&m.ticket, 1, Ordering.Relaxed);
+	for ticket != m.serving {
+		yield_processor();
+	}
+}
+
+ticket_mutex_unlock :: inline proc(m: ^Ticket_Mutex) {
+	atomic_add(&m.serving, 1, Ordering.Relaxed);
+}

core/sys/essence_linker_userland64.ld

@@ -0,0 +1,24 @@
+ENTRY(_start)
+
+SECTIONS
+{
+	. = 0x100000;
+	.text BLOCK(4K) : ALIGN(4K)
+	{
+		*(.text)
+	}
+	.rodata BLOCK(4K) : ALIGN(4K)
+	{
+		*(.rodata)
+	}
+	.data BLOCK(4K) : ALIGN(4K)
+	{
+		*(.data)
+	}
+
+	.bss BLOCK(4K) : ALIGN(4K)
+	{
+		*(COMMON)
+		*(.bss)
+	}
+}

core/sys/win32/comdlg32.odin

@@ -0,0 +1,189 @@
+// +build windows
+package win32
+
+foreign import "system:comdlg32.lib"
+import "core:strings"
+
+OFN_Hook_Proc :: #type proc "stdcall" (hdlg: Hwnd, msg: u32, wparam: Wparam, lparam: Lparam) -> Uint_Ptr;
+
+Open_File_Name_A :: struct {
+	struct_size:     u32,
+	hwnd_owner:      Hwnd,
+	instance:        Hinstance,
+	filter:          cstring,
+	custom_filter:   cstring,
+	max_cust_filter: u32,
+	filter_index:    u32,
+	file:            cstring,
+	max_file:        u32,
+	file_title:      cstring,
+	max_file_title:  u32,
+	initial_dir:     cstring,
+	title:           cstring,
+	flags:           u32,
+	file_offset:     u16,
+	file_extension:  u16,
+	def_ext:         cstring,
+	cust_data:       Lparam,
+	hook:            OFN_Hook_Proc,
+	template_name:   cstring,
+	pv_reserved:     rawptr,
+	dw_reserved:     u32,
+	flags_ex:        u32,
+}
+
+Open_File_Name_W :: struct {
+	struct_size:     u32,
+	hwnd_owner:      Hwnd,
+	instance:        Hinstance,
+	filter:          Wstring,
+	custom_filter:   Wstring,
+	max_cust_filter: u32,
+	filter_index:    u32,
+	file:            Wstring,
+	max_file:        u32,
+	file_title:      Wstring,
+	max_file_title:  u32,
+	initial_dir:     Wstring,
+	title:           Wstring,
+	flags:           u32,
+	file_offset:     u16,
+	file_extension:  u16,
+	def_ext:         Wstring,
+	cust_data:       Lparam,
+	hook:            OFN_Hook_Proc,
+	template_name:   Wstring,
+	pv_reserved:     rawptr,
+	dw_reserved:     u32,
+	flags_ex:        u32,
+}
+
+@(default_calling_convention = "c")
+foreign comdlg32 {
+	@(link_name="GetOpenFileNameA") get_open_file_name_a :: proc(arg1: ^Open_File_Name_A) -> Bool ---
+	@(link_name="GetOpenFileNameW") get_open_file_name_w :: proc(arg1: ^Open_File_Name_W) -> Bool ---
+	@(link_name="GetSaveFileNameA") get_save_file_name_a :: proc(arg1: ^Open_File_Name_A) -> Bool ---
+	@(link_name="GetSaveFileNameW") get_save_file_name_w :: proc(arg1: ^Open_File_Name_W) -> Bool ---
+	@(link_name="CommDlgExtendedError") comm_dlg_extended_error :: proc() -> u32 ---
+}
+
+OPEN_TITLE :: "Select file to open";
+OPEN_FLAGS :: u32(OFN_PATHMUSTEXIST | OFN_FILEMUSTEXIST);
+OPEN_FLAGS_MULTI :: u32(OPEN_FLAGS | OFN_ALLOWMULTISELECT | OFN_EXPLORER);
+
+SAVE_TITLE :: "Select file to save";
+SAVE_FLAGS :: u32(OFN_OVERWRITEPROMPT | OFN_EXPLORER);
+SAVE_EXT   :: "txt";
+
+Open_Save_Mode :: enum {
+	Open = 0,
+	Save = 1,
+}
+
+_open_file_dialog :: proc(title: string, dir: string,
+                          filters: []string, default_filter: u32,
+                          flags: u32, default_ext: string,
+                          mode: Open_Save_Mode, allocator := context.temp_allocator) -> (path: string, ok: bool = true) {
+	file_buf := make([]u16, MAX_PATH_WIDE, allocator);
+
+	// Filters need to be passed as a pair of strings (title, filter)
+	filter_len := u32(len(filters));
+	if filter_len % 2 != 0 do return "", false;
+	default_filter = clamp(default_filter, 1, filter_len / 2);
+
+	filter: string;
+	filter = strings.join(filters, "\u0000", context.temp_allocator);
+	filter = strings.concatenate({filter, "\u0000"}, context.temp_allocator);
+
+	ofn := Open_File_Name_W{
+		struct_size  = size_of(Open_File_Name_W),
+		file         = Wstring(&file_buf[0]),
+		max_file     = MAX_PATH_WIDE,
+		title        = utf8_to_wstring(title, context.temp_allocator),
+		filter       = utf8_to_wstring(filter, context.temp_allocator),
+		initial_dir  = utf8_to_wstring(dir, context.temp_allocator),
+		filter_index = u32(default_filter),
+		def_ext      = utf8_to_wstring(default_ext, context.temp_allocator),
+		flags        = u32(flags),
+	};
+
+	switch mode {
+	case .Open:
+		ok = bool(get_open_file_name_w(&ofn));
+	case .Save:
+		ok = bool(get_save_file_name_w(&ofn));
+	case:
+		ok = false;
+	}
+
+	if !ok {
+		delete(file_buf);
+		return "", false;
+	}
+
+	file_name := ucs2_to_utf8(file_buf[:], allocator);
+	path = strings.trim_right_null(file_name);
+	return;
+}
+
+select_file_to_open :: proc(title := OPEN_TITLE, dir := ".",
+                            filters := []string{"All Files", "*.*"}, default_filter := u32(1),
+                            flags := OPEN_FLAGS, allocator := context.temp_allocator) -> (path: string, ok: bool) {
+
+	path, ok = _open_file_dialog(title, dir, filters, default_filter, flags, "", Open_Save_Mode.Open, allocator);
+	return;
+}
+
+select_file_to_save :: proc(title := SAVE_TITLE, dir := ".",
+							filters := []string{"All Files", "*.*"}, default_filter := u32(1),
+							flags := SAVE_FLAGS, default_ext := SAVE_EXT,
+							allocator := context.temp_allocator) -> (path: string, ok: bool) {
+
+	path, ok = _open_file_dialog(title, dir, filters, default_filter, flags, default_ext, Open_Save_Mode.Save, allocator);
+	return;
+}
+
+// TODO: Implement convenience function for select_file_to_open with ALLOW_MULTI_SELECT that takes 
+//       it output of the form "path\u0000\file1u\0000file2" and turns it into []string with the path + file pre-concatenated for you.
+
+OFN_ALLOWMULTISELECT     :: 0x00000200; // NOTE(Jeroen): Without OFN_EXPLORER it uses the Win3 dialog.
+OFN_CREATEPROMPT         :: 0x00002000;
+OFN_DONTADDTORECENT      :: 0x02000000;
+OFN_ENABLEHOOK           :: 0x00000020;
+OFN_ENABLEINCLUDENOTIFY  :: 0x00400000;
+OFN_ENABLESIZING         :: 0x00800000;
+OFN_ENABLETEMPLATE       :: 0x00000040;
+OFN_ENABLETEMPLATEHANDLE :: 0x00000080;
+OFN_EXPLORER             :: 0x00080000;
+OFN_EXTENSIONDIFFERENT   :: 0x00000400;
+OFN_FILEMUSTEXIST        :: 0x00001000;
+OFN_FORCESHOWHIDDEN      :: 0x10000000;
+OFN_HIDEREADONLY         :: 0x00000004;
+OFN_LONGNAMES            :: 0x00200000;
+OFN_NOCHANGEDIR          :: 0x00000008;
+OFN_NODEREFERENCELINKS   :: 0x00100000;
+OFN_NOLONGNAMES          :: 0x00040000;
+OFN_NONETWORKBUTTON      :: 0x00020000;
+OFN_NOREADONLYRETURN     :: 0x00008000;
+OFN_NOTESTFILECREATE     :: 0x00010000;
+OFN_NOVALIDATE           :: 0x00000100;
+OFN_OVERWRITEPROMPT      :: 0x00000002;
+OFN_PATHMUSTEXIST        :: 0x00000800;
+OFN_READONLY             :: 0x00000001;
+OFN_SHAREAWARE           :: 0x00004000;
+OFN_SHOWHELP             :: 0x00000010;
+
+CDERR_DIALOGFAILURE      :: 0x0000FFFF;
+CDERR_GENERALCODES       :: 0x00000000;
+CDERR_STRUCTSIZE         :: 0x00000001;
+CDERR_INITIALIZATION     :: 0x00000002;
+CDERR_NOTEMPLATE         :: 0x00000003;
+CDERR_NOHINSTANCE        :: 0x00000004;
+CDERR_LOADSTRFAILURE     :: 0x00000005;
+CDERR_FINDRESFAILURE     :: 0x00000006;
+CDERR_LOADRESFAILURE     :: 0x00000007;
+CDERR_LOCKRESFAILURE     :: 0x00000008;
+CDERR_MEMALLOCFAILURE    :: 0x00000009;
+CDERR_MEMLOCKFAILURE     :: 0x0000000A;
+CDERR_NOHOOK             :: 0x0000000B;
+CDERR_REGISTERMSGFAIL    :: 0x0000000C;

core/sys/win32/crt.odin

@@ -0,0 +1,14 @@
+package win32
+
+import "core:strings";
+
+foreign {
+	@(link_name="_wgetcwd") _get_cwd_wide :: proc(buffer: Wstring, buf_len: int) ->  ^Wstring ---
+}
+
+get_cwd :: proc(allocator := context.temp_allocator) -> string {
+	buffer := make([]u16, MAX_PATH_WIDE, allocator);
+	_get_cwd_wide(Wstring(&buffer[0]), MAX_PATH_WIDE);
+	file := ucs2_to_utf8(buffer[:], allocator);
+	return strings.trim_right_null(file);
+}

core/sys/win32/gdi32.odin

@@ -0,0 +1,23 @@
+// +build windows
+package win32
+
+foreign import "system:gdi32.lib"
+
+
+@(default_calling_convention = "std")
+foreign gdi32 {
+	@(link_name="GetStockObject") get_stock_object :: proc(fn_object: i32) -> Hgdiobj ---;
+
+	@(link_name="StretchDIBits")
+	stretch_dibits :: proc(hdc: Hdc,
+	                       x_dst, y_dst, width_dst, height_dst: i32,
+	                       x_src, y_src, width_src, header_src: i32,
+	                       bits: rawptr, bits_info: ^Bitmap_Info,
+	                       usage: u32,
+	                       rop: u32) -> i32 ---;
+
+	@(link_name="SetPixelFormat")    set_pixel_format    :: proc(hdc: Hdc, pixel_format: i32, pfd: ^Pixel_Format_Descriptor) -> Bool ---;
+	@(link_name="ChoosePixelFormat") choose_pixel_format :: proc(hdc: Hdc, pfd: ^Pixel_Format_Descriptor) -> i32 ---;
+	@(link_name="SwapBuffers")       swap_buffers        :: proc(hdc: Hdc) -> Bool ---;
+
+}

core/sys/win32/helpers.odin

@@ -0,0 +1,29 @@
+// +build windows
+package win32
+
+import "core:strings";
+
+call_external_process :: proc(program, command_line: string) -> bool {
+    si := Startup_Info{ cb=size_of(Startup_Info) };
+    pi := Process_Information{};
+
+    return cast(bool)create_process_w(
+        utf8_to_wstring(program),
+        utf8_to_wstring(command_line),
+        nil, 
+        nil, 
+        Bool(false), 
+        u32(0x10), 
+        nil, 
+        nil, 
+        &si, 
+        &pi
+    );
+}
+
+open_website :: proc(url: string) -> bool {
+	p :: "C:\\Windows\\System32\\cmd.exe";
+	arg := []string{"/C", "start", url};
+	args := strings.join(arg, " ", context.temp_allocator);
+	return call_external_process(p, args);
+}

core/sys/win32/kernel32.odin

@@ -0,0 +1,221 @@
+// +build windows
+package win32
+
+foreign import "system:kernel32.lib"
+
+@(default_calling_convention = "std")
+foreign kernel32 {
+	@(link_name="CreateProcessA")		     create_process_a		      :: proc(application_name, command_line: cstring,
+	                                                              				 process_attributes, thread_attributes: ^Security_Attributes,
+	                                                              				 inherit_handle: Bool, creation_flags: u32, environment: rawptr,
+	                                                              				 current_direcotry: cstring, startup_info: ^Startup_Info,
+	                                                              				 process_information: ^Process_Information) -> Bool ---;
+    @(link_name="CreateProcessW")            create_process_w             :: proc(application_name, command_line: Wstring,
+                                                                                 process_attributes, thread_attributes: ^Security_Attributes,
+                                                                                 inherit_handle: Bool, creation_flags: u32, environment: rawptr,
+                                                                                 current_direcotry: cstring, startup_info: ^Startup_Info,
+                                                                                 process_information: ^Process_Information) -> Bool ---;
+	@(link_name="GetExitCodeProcess")		 get_exit_code_process        :: proc(process: Handle, exit: ^u32) -> Bool ---;
+	@(link_name="ExitProcess")               exit_process                 :: proc(exit_code: u32) ---;
+	@(link_name="GetModuleHandleA")          get_module_handle_a          :: proc(module_name: cstring) -> Hmodule ---;
+	@(link_name="GetModuleHandleW")          get_module_handle_w          :: proc(module_name: Wstring) -> Hmodule ---;
+
+	@(link_name="GetModuleFileNameA")        get_module_file_name_a       :: proc(module: Hmodule, filename: cstring, size: u32) -> u32 ---;
+	@(link_name="GetModuleFileNameW")        get_module_file_name_w       :: proc(module: Hmodule, filename: Wstring, size: u32) -> u32 ---;
+
+	@(link_name="Sleep")                     sleep                        :: proc(ms: i32) -> i32 ---;
+	@(link_name="QueryPerformanceFrequency") query_performance_frequency  :: proc(result: ^i64) -> i32 ---;
+	@(link_name="QueryPerformanceCounter")   query_performance_counter    :: proc(result: ^i64) -> i32 ---;
+	@(link_name="OutputDebugStringA")        output_debug_string_a        :: proc(c_str: cstring) ---;
+
+	@(link_name="GetCommandLineA")           get_command_line_a           :: proc() -> cstring ---;
+	@(link_name="GetCommandLineW")           get_command_line_w           :: proc() -> Wstring ---;
+	@(link_name="GetSystemMetrics")          get_system_metrics           :: proc(index: i32) -> i32 ---;
+	@(link_name="GetVersionExA")             get_version                  :: proc(osvi: ^OS_Version_Info_Ex_A) ---;
+	@(link_name="GetCurrentThreadId")        get_current_thread_id        :: proc() -> u32 ---;
+
+	@(link_name="GetSystemTimeAsFileTime")   get_system_time_as_file_time :: proc(system_time_as_file_time: ^Filetime) ---;
+	@(link_name="FileTimeToLocalFileTime")   file_time_to_local_file_time :: proc(file_time: ^Filetime, local_file_time: ^Filetime) -> Bool ---;
+	@(link_name="FileTimeToSystemTime")      file_time_to_system_time     :: proc(file_time: ^Filetime, system_time: ^Systemtime) -> Bool ---;
+	@(link_name="SystemTimeToFileTime")      system_time_to_file_time     :: proc(system_time: ^Systemtime, file_time: ^Filetime) -> Bool ---;
+
+	@(link_name="GetStdHandle")              get_std_handle               :: proc(h: i32) -> Handle ---;
+
+	@(link_name="CreateFileA")
+	create_file_a :: proc(filename: cstring, desired_access, share_module: u32,
+	                      security: rawptr,
+	                      creation, flags_and_attribs: u32, template_file: Handle) -> Handle ---;
+
+	@(link_name="CreateFileW")
+	create_file_w :: proc(filename: Wstring, desired_access, share_module: u32,
+	                      security: rawptr,
+	                      creation, flags_and_attribs: u32, template_file: Handle) -> Handle ---;
+
+
+	@(link_name="ReadFile")  read_file  :: proc(h: Handle, buf: rawptr, to_read: u32, bytes_read: ^i32, overlapped: rawptr) -> Bool ---;
+	@(link_name="WriteFile") write_file :: proc(h: Handle, buf: rawptr, len: i32, written_result: ^i32, overlapped: rawptr) -> Bool ---;
+
+	@(link_name="GetFileSizeEx")              get_file_size_ex               :: proc(file_handle: Handle, file_size: ^i64) -> Bool ---;
+	@(link_name="GetFileInformationByHandle") get_file_information_by_handle :: proc(file_handle: Handle, file_info: ^By_Handle_File_Information) -> Bool ---;
+
+	@(link_name="CreateDirectoryA") 		  create_directory_a			 :: proc(path: cstring, security_attributes: ^Security_Attributes) -> Bool ---;
+	@(link_name="CreateDirectoryW") 		  create_directory_w			 :: proc(path: Wstring, security_attributes: ^Security_Attributes) -> Bool ---;
+
+	@(link_name="GetFileType")    get_file_type    :: proc(file_handle: Handle) -> u32 ---;
+	@(link_name="SetFilePointer") set_file_pointer :: proc(file_handle: Handle, distance_to_move: i32, distance_to_move_high: ^i32, move_method: u32) -> u32 ---;
+
+	@(link_name="SetHandleInformation") set_handle_information :: proc(obj: Handle, mask, flags: u32) -> Bool ---;
+
+	@(link_name="FindFirstFileA") find_first_file_a :: proc(file_name: cstring, data: ^Find_Data_A) -> Handle ---;
+	@(link_name="FindNextFileA")  find_next_file_a  :: proc(file: Handle, data: ^Find_Data_A) -> Bool ---;
+
+	@(link_name="FindFirstFileW") find_first_file_w :: proc(file_name: Wstring, data: ^Find_Data_W) -> Handle ---;
+	@(link_name="FindNextFileW")  find_next_file_w  :: proc(file: Handle, data: ^Find_Data_W) -> Bool ---;
+
+	@(link_name="FindClose")      find_close        :: proc(file: Handle) -> Bool ---;
+
+	@(link_name="MoveFileExA")    move_file_ex_a    :: proc(existing, new: cstring, flags: u32) -> Bool ---;
+	@(link_name="DeleteFileA")    delete_file_a     :: proc(file_name: cstring) -> Bool ---;
+	@(link_name="CopyFileA")      copy_file_a       :: proc(existing, new: cstring, fail_if_exists: Bool) -> Bool ---;
+
+	@(link_name="MoveFileExW")    move_file_ex_w    :: proc(existing, new: Wstring, flags: u32) -> Bool ---;
+	@(link_name="DeleteFileW")    delete_file_w     :: proc(file_name: Wstring) -> Bool ---;
+	@(link_name="CopyFileW")      copy_file_w       :: proc(existing, new: Wstring, fail_if_exists: Bool) -> Bool ---;
+
+	@(link_name="HeapAlloc")      heap_alloc       :: proc(h: Handle, flags: u32, bytes: int) -> rawptr ---;
+	@(link_name="HeapReAlloc")    heap_realloc     :: proc(h: Handle, flags: u32, memory: rawptr, bytes: int) -> rawptr ---;
+	@(link_name="HeapFree")       heap_free        :: proc(h: Handle, flags: u32, memory: rawptr) -> Bool ---;
+	@(link_name="GetProcessHeap") get_process_heap :: proc() -> Handle ---;
+
+	@(link_name="LocalAlloc")     local_alloc      :: proc(flags: u32, bytes: int) -> rawptr ---;
+	@(link_name="LocalReAlloc")   local_realloc    :: proc(mem: rawptr, bytes: int, flags: uint) -> rawptr ---;
+	@(link_name="LocalFree")      local_free       :: proc(mem: rawptr) -> rawptr ---;
+
+	@(link_name="FindFirstChangeNotificationA") find_first_change_notification_a :: proc(path: cstring, watch_subtree: Bool, filter: u32) -> Handle ---;
+	@(link_name="FindNextChangeNotification")   find_next_change_notification    :: proc(h: Handle) -> Bool ---;
+	@(link_name="FindCloseChangeNotification")  find_close_change_notification   :: proc(h: Handle) -> Bool ---;
+
+	@(link_name="ReadDirectoryChangesW") read_directory_changes_w :: proc(dir: Handle, buf: rawptr, buf_length: u32,
+	                                                                      watch_subtree: Bool, notify_filter: u32,
+	                                                                      bytes_returned: ^u32, overlapped: ^Overlapped,
+	                                                                      completion: rawptr) -> Bool ---;
+
+	@(link_name="WideCharToMultiByte") wide_char_to_multi_byte :: proc(code_page: u32, flags: u32,
+	                                                                   wchar_str: Wstring, wchar: i32,
+	                                                                   multi_str: cstring, multi: i32,
+	                                                                   default_char: cstring, used_default_char: ^Bool) -> i32 ---;
+
+	@(link_name="MultiByteToWideChar") multi_byte_to_wide_char :: proc(code_page: u32, flags: u32,
+	                                                                   mb_str: cstring, mb: i32,
+	                                                                   wc_str: Wstring, wc: i32) -> i32 ---;
+
+	@(link_name="CreateSemaphoreA")    create_semaphore_a     :: proc(attributes: ^Security_Attributes, initial_count, maximum_count: i32, name: cstring) -> Handle ---;
+	@(link_name="CreateSemaphoreW")    create_semaphore_w     :: proc(attributes: ^Security_Attributes, initial_count, maximum_count: i32, name: cstring) -> Handle ---;
+	@(link_name="ReleaseSemaphore")    release_semaphore      :: proc(semaphore: Handle, release_count: i32, previous_count: ^i32) -> Bool ---;
+	@(link_name="WaitForSingleObject") wait_for_single_object :: proc(handle: Handle, milliseconds: u32) -> u32 ---;
+}
+
+// @(default_calling_convention = "c")
+foreign kernel32 {
+	@(link_name="GetLastError")              get_last_error               :: proc() -> i32 ---;
+	@(link_name="CloseHandle")               close_handle                 :: proc(h: Handle) -> i32 ---;
+
+	@(link_name="GetFileAttributesA")         get_file_attributes_a          :: proc(filename: cstring) -> u32 ---;
+	@(link_name="GetFileAttributesW")         get_file_attributes_w          :: proc(filename: Wstring) -> u32 ---;
+	@(link_name="GetFileAttributesExA")       get_file_attributes_ex_a       :: proc(filename: cstring, info_level_id: GET_FILEEX_INFO_LEVELS, file_info: ^File_Attribute_Data) -> Bool ---;
+	@(link_name="GetFileAttributesExW")       get_file_attributes_ex_w       :: proc(filename: Wstring, info_level_id: GET_FILEEX_INFO_LEVELS, file_info: ^File_Attribute_Data) -> Bool ---;
+	@(link_name="CompareFileTime")            compare_file_time              :: proc(a, b: ^Filetime) -> i32 ---;
+}
+
+@(default_calling_convention = "c")
+foreign kernel32 {
+	@(link_name="InterlockedCompareExchange") interlocked_compare_exchange :: proc(dst: ^i32, exchange, comparand: i32) -> i32 ---;
+	@(link_name="InterlockedExchange")        interlocked_exchange         :: proc(dst: ^i32, desired: i32) -> i32 ---;
+	@(link_name="InterlockedExchangeAdd")     interlocked_exchange_add     :: proc(dst: ^i32, desired: i32) -> i32 ---;
+	@(link_name="InterlockedAnd")             interlocked_and              :: proc(dst: ^i32, desired: i32) -> i32 ---;
+	@(link_name="InterlockedOr")              interlocked_or               :: proc(dst: ^i32, desired: i32) -> i32 ---;
+
+	@(link_name="InterlockedCompareExchange64") interlocked_compare_exchange64 :: proc(dst: ^i64, exchange, comparand: i64) -> i64 ---;
+	@(link_name="InterlockedExchange64")        interlocked_exchange64         :: proc(dst: ^i64, desired: i64) -> i64 ---;
+	@(link_name="InterlockedExchangeAdd64")     interlocked_exchange_add64     :: proc(dst: ^i64, desired: i64) -> i64 ---;
+	@(link_name="InterlockedAnd64")             interlocked_and64              :: proc(dst: ^i64, desired: i64) -> i64 ---;
+	@(link_name="InterlockedOr64")              interlocked_or64               :: proc(dst: ^i64, desired: i64) -> i64 ---;
+}
+
+@(default_calling_convention = "std")
+foreign kernel32 {
+	@(link_name="_mm_pause")        mm_pause           :: proc() ---;
+	@(link_name="ReadWriteBarrier") read_write_barrier :: proc() ---;
+	@(link_name="WriteBarrier")     write_barrier      :: proc() ---;
+	@(link_name="ReadBarrier")      read_barrier       :: proc() ---;
+
+	@(link_name="CreateThread")
+	create_thread :: proc(thread_attributes: ^Security_Attributes, stack_size: int, start_routine: rawptr,
+	                      parameter: rawptr, creation_flags: u32, thread_id: ^u32) -> Handle ---;
+	@(link_name="ResumeThread")      resume_thread        :: proc(thread: Handle) -> u32 ---;
+	@(link_name="GetThreadPriority") get_thread_priority  :: proc(thread: Handle) -> i32 ---;
+	@(link_name="SetThreadPriority") set_thread_priority  :: proc(thread: Handle, priority: i32) -> Bool ---;
+    @(link_name="GetExitCodeThread") get_exit_code_thread :: proc(thread: Handle, exit_code: ^u32) -> Bool ---;
+	@(link_name="TerminateThread")   terminate_thread     :: proc(thread: Handle, exit_code: u32) -> Bool ---;
+
+	@(link_name="InitializeCriticalSection")             initialize_critical_section                :: proc(critical_section: ^Critical_Section) ---;
+	@(link_name="InitializeCriticalSectionAndSpinCount") initialize_critical_section_and_spin_count :: proc(critical_section: ^Critical_Section, spin_count: u32) ---;
+	@(link_name="DeleteCriticalSection")                 delete_critical_section                    :: proc(critical_section: ^Critical_Section) ---;
+	@(link_name="SetCriticalSectionSpinCount")           set_critical_section_spin_count            :: proc(critical_section: ^Critical_Section, spin_count: u32) -> u32 ---;
+	@(link_name="TryEnterCriticalSection")               try_enter_critical_section                 :: proc(critical_section: ^Critical_Section) -> Bool ---;
+	@(link_name="EnterCriticalSection")                  enter_critical_section                     :: proc(critical_section: ^Critical_Section) ---;
+	@(link_name="LeaveCriticalSection")                  leave_critical_section                     :: proc(critical_section: ^Critical_Section) ---;
+
+	@(link_name="CreateEventA") create_event_a :: proc(event_attributes: ^Security_Attributes, manual_reset, initial_state: Bool, name: cstring) -> Handle ---;
+	@(link_name="CreateEventW") create_event_w :: proc(event_attributes: ^Security_Attributes, manual_reset, initial_state: Bool, name: Wstring) -> Handle ---;
+	@(link_name="PulseEvent")   pulse_event    :: proc(event: Handle) -> Bool ---;
+	@(link_name="SetEvent")     set_event      :: proc(event: Handle) -> Bool ---;
+	@(link_name="ResetEvent")   reset_event    :: proc(event: Handle) -> Bool ---;
+
+	@(link_name="LoadLibraryA")   load_library_a   :: proc(c_str: cstring)  -> Hmodule ---;
+	@(link_name="LoadLibraryW")   load_library_w   :: proc(c_str: Wstring) -> Hmodule ---;
+	@(link_name="FreeLibrary")    free_library     :: proc(h: Hmodule) -> Bool ---;
+	@(link_name="GetProcAddress") get_proc_address :: proc(h: Hmodule, c_str: cstring) -> rawptr ---;
+
+}
+
+Memory_Basic_Information :: struct {
+	base_address:       rawptr,
+	allocation_base:    rawptr,
+	allocation_protect: u32,
+	region_size:        uint,
+	state:              u32,
+	protect:            u32,
+	type:               u32,
+}
+
+@(default_calling_convention = "std")
+foreign kernel32 {
+	@(link_name="VirtualAlloc")   virtual_alloc    :: proc(address: rawptr, size: uint, allocation_type: u32, protect: u32) -> rawptr ---
+	@(link_name="VirtualAllocEx") virtual_alloc_ex :: proc(process: Handle, address: rawptr, size: uint, allocation_type: u32, protect: u32) -> rawptr ---
+	@(link_name="VirtualFree")    virtual_free     :: proc(address: rawptr, size: uint, free_type: u32) -> Bool ---
+	@(link_name="VirtualLock")    virtual_lock     :: proc(address: rawptr, size: uint) -> Bool ---
+	@(link_name="VirtualProtect") virtual_protect  :: proc(address: rawptr, size: uint, new_protect: u32, old_protect: ^u32) -> Bool ---
+	@(link_name="VirtualQuery")   virtual_query    :: proc(address: rawptr, buffer: ^Memory_Basic_Information, length: uint) -> uint ---
+}
+
+MEM_COMMIT      :: 0x00001000;
+MEM_RESERVE     :: 0x00002000;
+MEM_DECOMMIT    :: 0x00004000;
+MEM_RELEASE     :: 0x00008000;
+MEM_RESET       :: 0x00080000;
+MEM_RESET_UNDO  :: 0x01000000;
+
+MEM_LARGE_PAGES :: 0x20000000;
+MEM_PHYSICAL    :: 0x00400000;
+MEM_TOP_DOWN    :: 0x00100000;
+MEM_WRITE_WATCH :: 0x00200000;
+
+PAGE_NOACCESS           :: 0x01;
+PAGE_READONLY           :: 0x02;
+PAGE_READWRITE          :: 0x04;
+PAGE_WRITECOPY          :: 0x08;
+PAGE_EXECUTE            :: 0x10;
+PAGE_EXECUTE_READ       :: 0x20;
+PAGE_EXECUTE_READWRITE  :: 0x40;
+PAGE_EXECUTE_WRITECOPY  :: 0x80;

core/sys/win32/ole32.odin

@@ -0,0 +1,18 @@
+// +build windows
+package win32
+
+foreign import "system:ole32.lib"
+
+//objbase.h
+Com_Init :: enum {
+	Multi_Threaded = 0x0,
+	Apartment_Threaded = 0x2,
+	Disable_OLE1_DDE = 0x4,
+	Speed_Over_Memory = 0x8,
+};
+
+@(default_calling_convention = "std")
+foreign ole32 {
+	@(link_name ="CoInitializeEx") com_init_ex :: proc(reserved: rawptr, co_init: Com_Init) ->Hresult ---;
+	@(link_name = "CoUninitialize") com_shutdown :: proc() ---;
+}

core/sys/win32/shell32.odin

@@ -0,0 +1,9 @@
+// +build windows
+package win32
+
+foreign import "system:shell32.lib"
+
+@(default_calling_convention = "std")
+foreign shell32 {
+	@(link_name="CommandLineToArgvW") command_line_to_argv_w :: proc(cmd_list: Wstring, num_args: ^i32) -> ^Wstring ---;
+}

core/sys/win32/user32.odin

@@ -0,0 +1,270 @@
+// +build windows
+package win32
+
+foreign import "system:user32.lib"
+
+
+Menu_Bar_Info :: struct {
+	size: u32,
+	bar: Rect,
+	menu: Hmenu,
+	wnd_menu: Hwnd,
+	using fields: bit_field {
+		bar_focused: 1,
+		focuses:     1,
+	},
+}
+
+Menu_Item_Info_A :: struct {
+	size:          u32,
+	mask:          u32,
+	type:          u32,
+	state:         u32,
+	id:            u32,
+	submenu:       Hmenu,
+	bmp_checked:   Hbitmap,
+	bmp_unchecked: Hbitmap,
+	item_data:     u32,
+	type_data:     cstring,
+	cch:           u32,
+}
+Menu_Item_Info_W :: struct {
+	size:          u32,
+	mask:          u32,
+	type:          u32,
+	state:         u32,
+	id:            u32,
+	submenu:       Hmenu,
+	bmp_checked:   Hbitmap,
+	bmp_unchecked: Hbitmap,
+	item_data:     u32,
+	type_data:     Wstring,
+	cch:           u32,
+}
+
+MF_BYCOMMAND    :: 0x00000000;
+MF_BYPOSITION   :: 0x00000400;
+MF_BITMAP       :: 0x00000004;
+MF_CHECKED      :: 0x00000008;
+MF_DISABLED     :: 0x00000002;
+MF_ENABLED      :: 0x00000000;
+MF_GRAYED       :: 0x00000001;
+MF_MENUBARBREAK :: 0x00000020;
+MF_MENUBREAK    :: 0x00000040;
+MF_OWNERDRAW    :: 0x00000100;
+MF_POPUP        :: 0x00000010;
+MF_SEPARATOR    :: 0x00000800;
+MF_STRING       :: 0x00000000;
+MF_UNCHECKED    :: 0x00000000;
+
+MB_ABORTRETRYIGNORE     :: 0x00000002;
+MB_CANCELTRYCONTINUE    :: 0x00000006;
+MB_HELP                 :: 0x00004000;
+MB_OK                   :: 0x00000000;
+MB_OKCANCEL             :: 0x00000001;
+MB_RETRYCANCEL          :: 0x00000005;
+MB_YESNO                :: 0x00000004;
+MB_YESNOCANCEL          :: 0x00000003;
+
+MB_ICONEXCLAMATION      :: 0x00000030;
+MB_ICONWARNING          :: 0x00000030;
+MB_ICONINFORMATION      :: 0x00000040;
+MB_ICONASTERISK         :: 0x00000040;
+MB_ICONQUESTION         :: 0x00000020;
+MB_ICONSTOP             :: 0x00000010;
+MB_ICONERROR            :: 0x00000010;
+MB_ICONHAND             :: 0x00000010;
+
+MB_DEFBUTTON1           :: 0x00000000;
+MB_DEFBUTTON2           :: 0x00000100;
+MB_DEFBUTTON3           :: 0x00000200;
+MB_DEFBUTTON4           :: 0x00000300;
+
+MB_APPLMODAL            :: 0x00000000;
+MB_SYSTEMMODAL          :: 0x00001000;
+MB_TASKMODAL            :: 0x00002000;
+
+MB_DEFAULT_DESKTOP_ONLY :: 0x00020000;
+MB_RIGHT                :: 0x00080000;
+MB_RTLREADING           :: 0x00100000;
+MB_SETFOREGROUND        :: 0x00010000;
+MB_TOPMOST              :: 0x00040000;
+MB_SERVICE_NOTIFICATION :: 0x00200000;
+
+
+@(default_calling_convention = "std")
+foreign user32 {
+	@(link_name="GetDesktopWindow") get_desktop_window  :: proc() -> Hwnd ---;
+	@(link_name="ShowCursor")       show_cursor         :: proc(show: Bool) ---;
+	@(link_name="GetCursorPos")     get_cursor_pos      :: proc(p: ^Point) -> Bool ---;
+	@(link_name="SetCursorPos")     set_cursor_pos      :: proc(x, y: i32) -> Bool ---;
+	@(link_name="ScreenToClient")   screen_to_client    :: proc(h: Hwnd, p: ^Point) -> Bool ---;
+	@(link_name="ClientToScreen")   client_to_screen    :: proc(h: Hwnd, p: ^Point) -> Bool ---;
+	@(link_name="PostQuitMessage")  post_quit_message   :: proc(exit_code: i32) ---;
+	@(link_name="SetWindowTextA")   set_window_text_a   :: proc(hwnd: Hwnd, c_string: cstring) -> Bool ---;
+	@(link_name="SetWindowTextW")   set_window_text_w   :: proc(hwnd: Hwnd, c_string: Wstring) -> Bool ---;
+	@(link_name="RegisterClassExA") register_class_ex_a :: proc(wc: ^Wnd_Class_Ex_A) -> i16 ---;
+	@(link_name="RegisterClassExW") register_class_ex_w :: proc(wc: ^Wnd_Class_Ex_W) -> i16 ---;
+
+	@(link_name="CreateWindowExA")
+	create_window_ex_a :: proc(ex_style: u32,
+	                           class_name, title: cstring,
+	                           style: u32,
+	                           x, y, w, h: i32,
+	                           parent: Hwnd, menu: Hmenu, instance: Hinstance,
+	                           param: rawptr) -> Hwnd ---;
+
+	@(link_name="CreateWindowExW")
+	create_window_ex_w :: proc(ex_style: u32,
+	                           class_name, title: Wstring,
+	                           style: u32,
+	                           x, y, w, h: i32,
+	                           parent: Hwnd, menu: Hmenu, instance: Hinstance,
+	                           param: rawptr) -> Hwnd ---;
+
+	@(link_name="ShowWindow")       show_window        :: proc(hwnd: Hwnd, cmd_show: i32) -> Bool ---;
+	@(link_name="TranslateMessage") translate_message  :: proc(msg: ^Msg) -> Bool ---;
+	@(link_name="DispatchMessageA") dispatch_message_a :: proc(msg: ^Msg) -> Lresult ---;
+	@(link_name="DispatchMessageW") dispatch_message_w :: proc(msg: ^Msg) -> Lresult ---;
+	@(link_name="UpdateWindow")     update_window      :: proc(hwnd: Hwnd) -> Bool ---;
+	@(link_name="GetMessageA")      get_message_a      :: proc(msg: ^Msg, hwnd: Hwnd, msg_filter_min, msg_filter_max: u32) -> Bool ---;
+	@(link_name="GetMessageW")      get_message_w      :: proc(msg: ^Msg, hwnd: Hwnd, msg_filter_min, msg_filter_max: u32) -> Bool ---;
+
+	@(link_name="PeekMessageA") peek_message_a :: proc(msg: ^Msg, hwnd: Hwnd, msg_filter_min, msg_filter_max, remove_msg: u32) -> Bool ---;
+	@(link_name="PeekMessageW") peek_message_w :: proc(msg: ^Msg, hwnd: Hwnd, msg_filter_min, msg_filter_max, remove_msg: u32) -> Bool ---;
+
+
+	@(link_name="PostMessageA") post_message_a :: proc(hwnd: Hwnd, msg, wparam, lparam: u32) -> Bool ---;
+	@(link_name="PostMessageW") post_message_w :: proc(hwnd: Hwnd, msg, wparam, lparam: u32) -> Bool ---;
+
+	@(link_name="DefWindowProcA") def_window_proc_a :: proc(hwnd: Hwnd, msg: u32, wparam: Wparam, lparam: Lparam) -> Lresult ---;
+	@(link_name="DefWindowProcW") def_window_proc_w :: proc(hwnd: Hwnd, msg: u32, wparam: Wparam, lparam: Lparam) -> Lresult ---;
+
+	@(link_name="AdjustWindowRect") adjust_window_rect :: proc(rect: ^Rect, style: u32, menu: Bool) -> Bool ---;
+	@(link_name="GetActiveWindow")  get_active_window  :: proc() -> Hwnd ---;
+
+	@(link_name="DestroyWindow")       destroy_window        :: proc(wnd: Hwnd) -> Bool ---;
+	@(link_name="DescribePixelFormat") describe_pixel_format :: proc(dc: Hdc, pixel_format: i32, bytes: u32, pfd: ^Pixel_Format_Descriptor) -> i32 ---;
+
+	@(link_name="GetMonitorInfoA")  get_monitor_info_a  :: proc(monitor: Hmonitor, mi: ^Monitor_Info) -> Bool ---;
+	@(link_name="MonitorFromWindow") monitor_from_window :: proc(wnd: Hwnd, flags: u32) -> Hmonitor ---;
+
+	@(link_name="SetWindowPos") set_window_pos :: proc(wnd: Hwnd, wndInsertAfter: Hwnd, x, y, width, height: i32, flags: u32) ---;
+
+	@(link_name="GetWindowPlacement") get_window_placement  :: proc(wnd: Hwnd, wndpl: ^Window_Placement) -> Bool ---;
+	@(link_name="SetWindowPlacement") set_window_placement  :: proc(wnd: Hwnd, wndpl: ^Window_Placement) -> Bool ---;
+	@(link_name="GetWindowRect")      get_window_rect       :: proc(wnd: Hwnd, rect: ^Rect) -> Bool ---;
+
+	@(link_name="GetWindowLongPtrA") get_window_long_ptr_a :: proc(wnd: Hwnd, index: i32) -> Long_Ptr ---;
+	@(link_name="SetWindowLongPtrA") set_window_long_ptr_a :: proc(wnd: Hwnd, index: i32, new: Long_Ptr) -> Long_Ptr ---;
+	@(link_name="GetWindowLongPtrW") get_window_long_ptr_w :: proc(wnd: Hwnd, index: i32) -> Long_Ptr ---;
+	@(link_name="SetWindowLongPtrW") set_window_long_ptr_w :: proc(wnd: Hwnd, index: i32, new: Long_Ptr) -> Long_Ptr ---;
+
+	@(link_name="GetWindowText") get_window_text :: proc(wnd: Hwnd, str: cstring, maxCount: i32) -> i32 ---;
+
+	@(link_name="GetClientRect") get_client_rect :: proc(hwnd: Hwnd, rect: ^Rect) -> Bool ---;
+
+	@(link_name="GetDC")     get_dc     :: proc(h: Hwnd) -> Hdc ---;
+	@(link_name="ReleaseDC") release_dc :: proc(wnd: Hwnd, hdc: Hdc) -> i32 ---;
+
+	@(link_name="MapVirtualKeyA") map_virtual_key_a :: proc(scancode: u32, map_type: u32) -> u32 ---;
+	@(link_name="MapVirtualKeyW") map_virtual_key_w :: proc(scancode: u32, map_type: u32) -> u32 ---;
+
+	@(link_name="GetKeyState")      get_key_state       :: proc(v_key: i32) -> i16 ---;
+	@(link_name="GetAsyncKeyState") get_async_key_state :: proc(v_key: i32) -> i16 ---;
+
+	@(link_name="SetForegroundWindow") set_foreground_window :: proc(h: Hwnd) -> Bool ---;
+	@(link_name="SetFocus")            set_focus             :: proc(h: Hwnd) -> Hwnd ---;
+
+
+    @(link_name="LoadImageA")       load_image_a        :: proc(instance: Hinstance, name: cstring, type_: u32, x_desired, y_desired : i32, load : u32) -> Handle ---;
+    @(link_name="LoadIconA")        load_icon_a         :: proc(instance: Hinstance, icon_name: cstring) -> Hicon ---;
+    @(link_name="DestroyIcon")      destroy_icon        :: proc(icon: Hicon) -> Bool ---;
+
+    @(link_name="LoadCursorA")      load_cursor_a       :: proc(instance: Hinstance, cursor_name: cstring) -> Hcursor ---;
+    @(link_name="LoadCursorW")      load_cursor_w       :: proc(instance: Hinstance, cursor_name: cstring) -> Hcursor ---;
+	@(link_name="GetCursor")        get_cursor          :: proc() -> Hcursor ---;
+	@(link_name="SetCursor")        set_cursor          :: proc(cursor: Hcursor) -> Hcursor ---;
+
+	@(link_name="RegisterRawInputDevices") register_raw_input_devices :: proc(raw_input_device: ^Raw_Input_Device, num_devices, size: u32) -> Bool ---;
+
+	@(link_name="GetRawInputData") get_raw_input_data :: proc(raw_input: Hrawinput, command: u32, data: rawptr, size: ^u32, size_header: u32) -> u32 ---;
+
+	@(link_name="MapVirtualKeyExW") map_virtual_key_ex_w :: proc(code, map_type: u32, hkl: HKL) ---;
+	@(link_name="MapVirtualKeyExA") map_virtual_key_ex_a :: proc(code, map_type: u32, hkl: HKL) ---;
+
+	@(link_name="EnumDisplayMonitors") enum_display_monitors :: proc(hdc: Hdc,  rect: ^Rect, enum_proc: Monitor_Enum_Proc, lparam: Lparam) -> bool ---;
+}
+
+@(default_calling_convention = "c")
+foreign user32 {
+	@(link_name="CreateMenu")      create_menu   :: proc() -> Hmenu ---
+	@(link_name="CreatePopupMenu") create_popup_menu :: proc() -> Hmenu ---
+	@(link_name="DestroyMenu")     destroy_menu :: proc(menu: Hmenu) -> Bool ---
+	@(link_name="DeleteMenu")      delete_menu :: proc(menu: Hmenu, position: u32, flags: u32) -> Bool ---
+
+	@(link_name="EnableMenuItem")  enable_menu_item :: proc(menu: Hmenu, id_enable_itme: i32, enable: u32) -> Bool ---
+	@(link_name="EndMenu")         end_menu :: proc() -> Bool ---
+	@(link_name="GetMenu")         get_menu :: proc(wnd: Hwnd) -> Hmenu ---
+	@(link_name="GetMenuBarInfo")  get_menu_bar_info :: proc(wnd: Hwnd, id_object, id_item: u32, mbi: ^Menu_Bar_Info) -> Hmenu ---
+	@(link_name="GetMenuStringA")  get_menu_string_a :: proc(menu: Hmenu, id_item: u32, s: string,  cch_max: i32, flags: u32) -> i32 ---
+	@(link_name="GetMenuStringW")  get_menu_string_w :: proc(menu: Hmenu, id_item: u32, s: Wstring, cch_max: i32, flags: u32) -> i32 ---
+	@(link_name="GetMenuState")    get_menu_state :: proc(menu: Hmenu, id: u32, flags: u32) -> u32 ---
+	@(link_name="GetMenuItemRect") get_menu_item_rect :: proc(wnd: Hwnd, menu: Hmenu, id_item: u32, item: ^Rect) -> Bool ---
+
+	@(link_name="SetMenu")         set_menu :: proc(wnd: Hwnd, menu: Hmenu) -> Hmenu ---
+
+	@(link_name="DrawMenuBar")     draw_menu_bar :: proc(wnd: Hwnd) -> Bool ---
+	@(link_name="InsertMenuA")     insert_menu_a :: proc(menu: Hmenu, position: u32, flags: u32, id_new_item: Uint_Ptr, new_item: cstring) -> Bool ---
+	@(link_name="InsertMenuW")     insert_menu_w :: proc(menu: Hmenu, position: u32, flags: u32, id_new_item: Uint_Ptr, new_item: Wstring) -> Bool ---
+
+	@(link_name="InsertMenuItemA") insert_menu_item_a :: proc(menu: Hmenu, item: u32, by_position: bool, mi: ^Menu_Item_Info_A) -> Bool ---
+	@(link_name="InsertMenuItemW") insert_menu_item_w :: proc(menu: Hmenu, item: u32, by_position: bool, mi: ^Menu_Item_Info_W) -> Bool ---
+
+	@(link_name="AppendMenuA") append_menu_a :: proc(menu: Hmenu, flags: u32, id_new_item: Uint_Ptr, new_item: cstring) -> Bool ---
+	@(link_name="AppendMenuW") append_menu_w :: proc(menu: Hmenu, flags: u32, id_new_item: Uint_Ptr, new_item: Wstring) -> Bool ---
+
+	@(link_name="CheckMenuItem") check_menu_item :: proc(menu: Hmenu, id_check_item: u32, check: u32) -> u32 ---
+	@(link_name="CheckMenuRadioItem") check_menu_radio_item :: proc(menu: Hmenu, first, last: u32, check: u32, flags: u32) -> Bool ---
+
+	@(link_name="GetPropA") get_prop_a :: proc(wnd: Hwnd, s: cstring) -> Handle ---
+	@(link_name="GetPropW") get_prop_w :: proc(wnd: Hwnd, s: Wstring) -> Handle ---
+
+
+	@(link_name="MessageBoxExA") message_box_ex_a :: proc(wnd: Hwnd, text, caption: cstring, type: u32, language_id: u16) -> i32 ---
+	@(link_name="MessageBoxExW") message_box_ex_w :: proc(wnd: Hwnd, text, caption: Wstring, type: u32, language_id: u16) -> i32 ---
+}
+
+
+_IDC_APPSTARTING := rawptr(uintptr(32650));
+_IDC_ARROW       := rawptr(uintptr(32512));
+_IDC_CROSS       := rawptr(uintptr(32515));
+_IDC_HAND        := rawptr(uintptr(32649));
+_IDC_HELP        := rawptr(uintptr(32651));
+_IDC_IBEAM       := rawptr(uintptr(32513));
+_IDC_ICON        := rawptr(uintptr(32641));
+_IDC_NO          := rawptr(uintptr(32648));
+_IDC_SIZE        := rawptr(uintptr(32640));
+_IDC_SIZEALL     := rawptr(uintptr(32646));
+_IDC_SIZENESW    := rawptr(uintptr(32643));
+_IDC_SIZENS      := rawptr(uintptr(32645));
+_IDC_SIZENWSE    := rawptr(uintptr(32642));
+_IDC_SIZEWE      := rawptr(uintptr(32644));
+_IDC_UPARROW     := rawptr(uintptr(32516));
+_IDC_WAIT        := rawptr(uintptr(32514));
+IDC_APPSTARTING := cstring(_IDC_APPSTARTING);
+IDC_ARROW       := cstring(_IDC_ARROW);
+IDC_CROSS       := cstring(_IDC_CROSS);
+IDC_HAND        := cstring(_IDC_HAND);
+IDC_HELP        := cstring(_IDC_HELP);
+IDC_IBEAM       := cstring(_IDC_IBEAM);
+IDC_ICON        := cstring(_IDC_ICON);
+IDC_NO          := cstring(_IDC_NO);
+IDC_SIZE        := cstring(_IDC_SIZE);
+IDC_SIZEALL     := cstring(_IDC_SIZEALL);
+IDC_SIZENESW    := cstring(_IDC_SIZENESW);
+IDC_SIZENS      := cstring(_IDC_SIZENS);
+IDC_SIZENWSE    := cstring(_IDC_SIZENWSE);
+IDC_SIZEWE      := cstring(_IDC_SIZEWE);
+IDC_UPARROW     := cstring(_IDC_UPARROW);
+IDC_WAIT        := cstring(_IDC_WAIT);

core/sys/win32/wgl.odin

@@ -0,0 +1,114 @@
+// +build windows
+package win32
+
+foreign import "system:opengl32.lib"
+
+CONTEXT_MAJOR_VERSION_ARB             :: 0x2091;
+CONTEXT_MINOR_VERSION_ARB             :: 0x2092;
+CONTEXT_FLAGS_ARB                     :: 0x2094;
+CONTEXT_PROFILE_MASK_ARB              :: 0x9126;
+CONTEXT_FORWARD_COMPATIBLE_BIT_ARB    :: 0x0002;
+CONTEXT_CORE_PROFILE_BIT_ARB          :: 0x00000001;
+CONTEXT_COMPATIBILITY_PROFILE_BIT_ARB :: 0x00000002;
+
+Hglrc     :: distinct Handle;
+Color_Ref :: distinct u32;
+
+Layer_Plane_Descriptor :: struct {
+	size:             u16,
+	version:          u16,
+	flags:            u32,
+	pixel_type:       u8,
+	color_bits:       u8,
+	red_bits:         u8,
+	red_shift:        u8,
+	green_bits:       u8,
+	green_shift:      u8,
+	blue_bits:        u8,
+	blue_shift:       u8,
+	alpha_bits:       u8,
+	alpha_shift:      u8,
+	accum_bits:       u8,
+	accum_red_bits:   u8,
+	accum_green_bits: u8,
+	accum_blue_bits:  u8,
+	accum_alpha_bits: u8,
+	depth_bits:       u8,
+	stencil_bits:     u8,
+	aux_buffers:      u8,
+	layer_type:       u8,
+	reserved:         u8,
+	transparent:      Color_Ref,
+}
+
+Point_Float :: struct {x, y: f32};
+
+Glyph_Metrics_Float :: struct {
+	black_box_x:  f32,
+	black_box_y:  f32,
+	glyph_origin: Point_Float,
+	cell_inc_x:   f32,
+	cell_inc_y:   f32,
+}
+
+Create_Context_Attribs_ARB_Type :: #type proc "c" (hdc: Hdc, h_share_context: rawptr, attribList: ^i32) -> Hglrc;
+Choose_Pixel_Format_ARB_Type    :: #type proc "c" (hdc: Hdc, attrib_i_list: ^i32, attrib_f_list: ^f32, max_formats: u32, formats: ^i32, num_formats : ^u32) -> Bool;
+Swap_Interval_EXT_Type          :: #type proc "c" (interval: i32) -> bool;
+Get_Extensions_String_ARB_Type  :: #type proc "c" (Hdc) -> cstring;
+
+// Procedures
+	create_context_attribs_arb: Create_Context_Attribs_ARB_Type;
+	choose_pixel_format_arb:    Choose_Pixel_Format_ARB_Type;
+	swap_interval_ext:          Swap_Interval_EXT_Type;
+	get_extensions_string_arb:  Get_Extensions_String_ARB_Type;
+
+
+foreign opengl32 {
+	@(link_name="wglCreateContext")
+	create_context :: proc(hdc: Hdc) -> Hglrc ---;
+
+	@(link_name="wglMakeCurrent")
+	make_current :: proc(hdc: Hdc, hglrc: Hglrc) -> Bool ---;
+
+	@(link_name="wglGetProcAddress")
+	get_gl_proc_address :: proc(c_str: cstring) -> rawptr ---;
+
+	@(link_name="wglDeleteContext")
+	delete_context :: proc(hglrc: Hglrc) -> Bool ---;
+
+	@(link_name="wglCopyContext")
+	copy_context :: proc(src, dst: Hglrc, mask: u32) -> Bool ---;
+
+	@(link_name="wglCreateLayerContext")
+	create_layer_context :: proc(hdc: Hdc, layer_plane: i32) -> Hglrc ---;
+
+	@(link_name="wglDescribeLayerPlane")
+	describe_layer_plane :: proc(hdc: Hdc, pixel_format, layer_plane: i32, bytes: u32, pd: ^Layer_Plane_Descriptor) -> Bool ---;
+
+	@(link_name="wglGetCurrentContext")
+	get_current_context :: proc() -> Hglrc ---;
+
+	@(link_name="wglGetCurrentDC")
+	get_current_dc :: proc() -> Hdc ---;
+
+	@(link_name="wglGetLayerPaletteEntries")
+	get_layer_palette_entries :: proc(hdc: Hdc, layer_plane, start, entries: i32, cr: ^Color_Ref) -> i32 ---;
+
+	@(link_name="wglRealizeLayerPalette")
+	realize_layer_palette :: proc(hdc: Hdc, layer_plane: i32, realize: Bool) -> Bool ---;
+
+	@(link_name="wglSetLayerPaletteEntries")
+	set_layer_palette_entries :: proc(hdc: Hdc, layer_plane, start, entries: i32, cr: ^Color_Ref) -> i32 ---;
+
+	@(link_name="wglShareLists")
+	share_lists :: proc(hglrc1, hglrc2: Hglrc) -> Bool ---;
+
+	@(link_name="wglSwapLayerBuffers")
+	swap_layer_buffers :: proc(hdc: Hdc, planes: u32) -> Bool ---;
+
+	@(link_name="wglUseFontBitmaps")
+	use_font_bitmaps :: proc(hdc: Hdc, first, count, list_base: u32) -> Bool ---;
+
+	@(link_name="wglUseFontOutlines")
+	use_font_outlines :: proc(hdc: Hdc, first, count, list_base: u32, deviation, extrusion: f32, format: i32, gmf: ^Glyph_Metrics_Float) -> Bool ---;
+}

core/sys/win32/winmm.odin

@@ -0,0 +1,10 @@
+// +build windows
+package win32
+
+foreign import "system:winmm.lib"
+
+
+@(default_calling_convention = "std")
+foreign winmm {
+	@(link_name="timeGetTime") time_get_time :: proc() -> u32 ---;
+}